Six soybean populations were developed, utilizing the Sclerotinia stem rot resistance sources AxN-1-55 and W04-1002 (Table 1). AxN-1-55 (PI 640911) was released as public germplasm (Diers et al., 2006) with partial Sclerotinia stem rot resistance in 2006. AxN-1-55 has 75% plant survival after challenge with S. sclerotiorum in controlled inoculation trials (Grau unpublished data). W04-1002 is an inbred line derived from a single plant selection from PI 567157A (Peltier and Grau, 2008). W04-1002 has expressed 90 to 100% survival after repeated challenges with multiple isolates of S. sclerotiorum and is considered highly resistant to the pathogen (Peltier and Grau, 2008).

The two aforementioned sources of resistance to S. sclerotiorum were selected as parents to cross with four parental lines possessing desirable agronomic and pathogen resistance traits (Table 1). Populations were assigned a name containing a number indicating the female parent and a “1” or “2” for the Sclerotinia stem rot resistant parent, W04-1002 or AxN-1-55, respectively. Population designations are as follows: 41 = W04-571 × W04-1002, 51 = W04-680 × W04-1002, 81 = L84-5873 × W04-1002, 91 = LN89–5717 × W04-1002, 42 = W04-571 × AxN-1-55, 52 = W04-680 × AxN-1-55.

Initial selections were made based on pod set, minimal lodging, maturity (MG0 to MGII), and absence of foliar diseases to ensure the persistence of acceptable agronomic qualities and parental C. gregata and H. glycines resistance. Six F1 seeds from each of the six populations were planted in a greenhouse; F2 seed was harvested and combined for field selection. In 2007 F2 seed was planted in a field nursery, naturally infested with C. gregata and H. glycines, at West Madison Agricultural Research Station located in Verona, Wisconsin (43.06028, -89.531667). Approximately 300 plants were selected within each population, harvested, and F3 seed was combined for planting in 2008. In 2008 field selections, the identity of progeny of each selected plant was maintained to develop sets of individual F3:4 breeding lines for each population.

Selection for resistance to S. sclerotiorum was conducted in greenhouse studies and in naturally infested field nurseries. Four plants per F3:4 line were challenged with S. sclerotiorum isolate 105HT (Peltier and Grau, 2008) in greenhouse trials to select for physiological resistance. A cut-petiole inoculation technique was used to challenge lines at the R1 (first flower) growth stage (Peltier and Grau, 2008). One or two surviving plants were advanced to the next generation of selection and identified as a new line. Remnant seed from lines designated as susceptible (plants from lines with 100% mortality when inoculated) were planted to maintain both resistant and susceptible lines within each population in order to assess genetic gain from selection. Single plant selection of new lines continued until the F7 generation, and 1,076 F7:8 lines were advanced to the field for further selection.

After three generations of greenhouse selection, 1,076 inbred lines (F7:8) were planted in 6.1 m, single row, non-replicated plots in a field nursery naturally infested with C. gregata and H. glycines. Eight hundred and thirty lines were selected for the persistence of acceptable agronomic traits and disease resistance based on criteria previously described. All breeding lines (F7:9 generation) within the four W04-1002-descended populations were once again evaluated for resistance to S. sclerotiorum in greenhouse trials. Selection within population 42 was discontinued due to a lack of sustained and measurable disease resistance. By the end of this selection phase of the project, there were 109 lines for population 41, 117 lines for population 51, 224 lines for population 81 and 250 lines for population 91 for a total of 700 lines.

Greenhouse trials were conducted in 2013 at the West Madison Greenhouse Complex located on the grounds of West Madison Agricultural Research Station. Soybean seeds were planted approximately 4 cm deep in 15.25 cm diameter pots of moist potting mix (Sun Gro Horticulture). Soybean plants were watered daily and fertilized twice weekly (Scotts Peters Professional Peat-Lite Special 20-10-20; Scotts-Sierra Horticultural Products Co.) prior to inoculation.

Soybean plants were inoculated using the cut petiole technique (Peltier and Grau, 2008) with aggressive S. sclerotiorum isolate 25 (Willbur et al., 2017). A 1.5-cm-thick agar core was collected from the leading edge of mycelia on each inoculum plate with a 1,000 μl pipet tip (Fisher Scientific). At the R1 (first flower) growth stage, second or third trifoliate leaflets were excised at a petiole length of 2.5 to 3 cm. Pipet tips of inoculum were placed on petioles such that mycelia and cut petiole tissue were in direct contact. Two to three plants (sub-samples) were inoculated per pot for each line and replicated three to four times in a randomized complete block design (RCBD) blocked by replicate. The trial was repeated once. F7:9 lines were phenotyped at the R3 growth stage 14 days post-inoculation (DPI) for resistance to Sclerotinia stem rot using a rating scale of 0 (no stem lesion), 1 (small stem lesion), 2 (lesion but no wilt), 3 (wilt), and 4 (dead plant). Lines with a mean severity score of 0 to 1 were characterized as resistant.

This work focused on mapping genes that control Sclerotinia stem rot infection using the W04-1002 lines as a resistance source because it represents a novel and stable source of Sclerotinia stem rot resistance. Genomic DNA was extracted from a bulk sample of fresh leaf tissue from each of the 8–10 most Sclerotinia stem rot-resistant and most Sclerotinia stem rot-susceptible lines in 2013 greenhouse evaluations (Table 2). Seven soybean leaves for each line were used for the hexadecyltrimethylammonium bromide (CTAB) extraction protocol as described by Keim et al. (1988). The samples were tested for single nucleotide polymorphism (SNP) genetic markers with the Illumina GoldenGate 1,536 Universal Soy Linkage Panel 1.0 (USLP 1.0) (Hyten et al., 2010). Marker data were analyzed for an association between disease resistance and the alleles from the resistant and susceptible parents for each marker. Based on chi-square analysis, SNP markers with significant segregation distortion were identified. Subsequently, microsatellite markers closely linked to the significant SNP markers from the Chi-square analysis were used to evaluate all lines in the populations. Primer sequences for microsatellite markers were obtained from the SoyBase website¹. Genetic markers were evaluated for 109 lines in population 41, 117 lines in population 51, 224 lines in population 81 and 250 lines in population 91, for a total of 700 lines evaluated. Lines were tested with 12 to 37 markers, depending on the population, that mapped onto three to seven chromosomes (Table 3).

Lines planted in the 2014 advanced field trials were selected based on the lowest Sclerotinia stem rot disease severity of the 336 lines trialed from the six populations in the 2013 naturally infested field trials (population 41 = 64 lines; population 51 = 51 lines; population 81 = 93 lines; population 91 = 76 lines; population 42 = 36 lines; population 52 = 16 lines). Subsequent evaluations were also performed in the greenhouse in early 2014. As mentioned previously, population 42 was not evaluated through 2015 due to a lack of performance.

The lines selected in 2013 were planted at the West Madison Agricultural Research Station in May of 2014 and the Hancock Agricultural Research Station in May of 2015 and 2016. Plots were overhead irrigated at each location 1.9–3.2 cm/ha every 2–5 days to facilitate disease development.

The experimental design each year was a randomized complete block (blocked by replicate) with five or six replications. Plots consisted of four 0.76 m wide rows that were 6.1 m long. Each plot was separated by a 1.5 m non-planted alley. Sowing occurred at a rate of approximately 437,500 seeds ha^-1 using a tractor-mounted cone-type planter (Almaco, Nevada, IA). Nutrient management was conducted per University of Wisconsin-Madison Cropping Guidelines.

Plot grain weight and moisture was taken from the two center rows of each field plot using an Almaco (Nevada, IA, United States) SPC40 small-plot combine equipped with a HarvestMaster HM800 grain gauge with Mirus software package (Juniper Systems, Logan, UT, United States). Yield measurements were calculated and standardized to 13% moisture. Sub-samples of grain were obtained during harvest and used to assess oil (%) and protein (%) in 2015 and 2016. Oil and protein data were assessed using the average of five 50 ml subsamples of seed from each plot using a near infrared (NIR) grain analyzer (Perten Instruments Inframatic 9500, Hägersten, Sweden). Readings were calibrated by the system for a moisture content of 13%.

Sclerotinia stem rot severity index (DSI) was determined in all years by a rating 30 arbitrarily selected plants in each plot of the field nursery at the R6 soybean growth stage. Plants were scored either 0 (no infection), 1 (infection on branches), 2 (infection on, but not girdling, the main stem), or 3 (infection on the main stem resulting in death or poor pod fill). The sum of the scores of the 30 plants were totaled for each class and divided by 0.9 (Grau et al., 1982). The disease incidence (DI) was calculated by counting the number of symptomatic plants in 12.19 m of row. Lodging was measured October 11, 2014, October 10, 2015, and October 14, 2016 using an average ranking for each plot of 1 (no leaning), 2 (25-degree lean), 3 (45-degree lean), 4 (more than a 45-degree lean), and 5 (laying on the ground) for each plot.

Multi-isolate greenhouse evaluations were conducted on soybean plants from the seed of lines tested in 2016 using the previously described cut-petiole technique (Peltier and Grau, 2008). Three soybean seeds per pot were planted approximately 4 cm deep in 15.25 cm diameter peat pots of moist potting mix (Premier Pro-Mix HP BioFungicide + Mycorrhizae). Nine S. sclerotiorum isolates, of 44 previously characterized isolates with varying degrees of aggressiveness (Willbur et al., 2017), were used for evaluations. As described by Willbur et al. (2017), lines were inoculated at the V4 growth stage, and lesions were measured with digital calipers (Thermo Fisher Scientific) 5, 12, and 14 DPI. Inoculations were performed in triplicate (three replicates) with three seeds planted per pot, and soybean plants were arranged in a RCBD blocked by replicate. Line evaluations were repeated a second time. Initial inoculum was generated from dry-stored sclerotia (Willbur et al., 2017). Inoculum applied in the second repetition was generated from sclerotia reisolated from the first repetition of plants. The AUDPC was analyzed to evaluate germplasm resistance reactions to a variety of isolates in both greenhouse screens.

Mixed-model analysis of variance (ANOVA) was conducted for lodging, agronomic traits, and protein and oil using PROC GLIMMIX in the SAS statistical software package and the analysis of the markers tested for populations was conducted in PROC GLM (v 9.4, SAS Institute, Inc. Cary, NC, United States). Means were separated using Fisher’s Least Significant Difference (LSD) via an open source macro (Piepho, 2012). Prior to analysis, lodging scores for each plot were subjected to rank analysis using PROC RANK in SAS. This was done to normalize the categorical nature of lodging scores, so that mixed model ANOVA could be conducted as described above. Disease and yield data were analyzed separately for each year due to large differences in overall disease attributable to environment variability. Significance was reported at α = 0.05 significance level.

The multi-isolate-germplasm line experiments were analyzed using a generalized mixed model (PROC GLIMMIX) analysis of variance using SAS (v 9.4, SAS Institute, Inc.), as described in Willbur et al. (2017). Data were normalized using a lognormal distribution and denominator degrees of freedom for fixed effects were computed using the Kenward-Rodger degrees of freedom approximation. Differences between lines and isolates were determined at α = 0.05 significance level.

Six populations were generated in this study by utilizing two Sclerotinia stem rot resistance sources AxN-1-55 and W04-1002. Crosses were established between these sources of resistance and six parental lines conferring other desirable pathogen resistance traits (See materials and methods for details). The resulting populations were designated as follows: 41 = W04-571 × W04-1002, 51 = W04-680 × W04-1002, 81 = L84–5873 × W04-1002, 91 = LN89–5717 × W04-1002, 42 = W04-571 × AxN-1-55, 52 = W04-680 × AxN-1-55. In 2013, 700 promising inbred lines (F7:9) derived from W04-1002, were subjected to S. sclerotiorum petiole inoculations to evaluate physiological resistance to S. sclerotiorum and to later identify markers associated with resistance phenotypes. Responses observed among the lines ranged from resistant to highly susceptible within greenhouse trials (Figure 1). After multiple greenhouse trials, 160 of 700 expressed 0 to 25% plant mortality (severity class 0–1, Figure 1). Concurrently, parental lines were also evaluated for Sclerotinia stem rot resistance. As expected, the resistant parent W04-1002 showed a highly resistant rating of 0.1, while the average rating of W04-571 was 3.0, W04-680 was 4.0, L84–5873 was 3.5, and LN89–5717 was 3.8. W04-1002, therefore, remained one of the most resistant lines, and a range of responses to Sclerotinia stem rot persisted among lines at the F7:9 generation. The results of 2013 greenhouse evaluations informed selection of the most resistant and susceptible lines for SNP analyses in 2013 and the selection W04-1002 lines evaluated in 2014 field trials.

The preliminary marker analysis was performed using 1,536 SNP genetic markers (data not shown). This was done by comparing the marker pattern of the 8–10 most resistant lines with the 8–10 most susceptible lines in each population generated from a cross with W04-1002 (Table 2). The SNP markers data were analyzed to determine if there was an association between disease resistance and the alleles from the resistant and susceptible parents for each marker. This association was determined by testing for significant segregation distortion compared to expected random segregation. A significant distortion of segregation indicated that the marker was genetically close to a resistance allele. Based on chi-square analysis, markers were identified that had significant (P < 0.05) segregation distortion. Due to the high cost of testing all lines in the four populations with the 1,536 SNP markers, microsatellite markers in regions where the distorted markers are located were then used to test all lines in the populations (Table 3).

The microsatellite marker results from all lines in the populations were then combined with the resistance data from the 2013 greenhouse evaluations to map QTL in each population. Using a significance threshold of P < 0.05, this analysis resulted in the mapping of QTL to one region on chromosome 15 in population 41, one region on chromosome 19 in population 51, regions on chromosomes 17 and 18 in population 81, and regions on chromosomes 16 and 19 in population 91 (Table 4). For many of these regions, multiple significant markers were identified in each population, but these areas where multiple markers arise are likely linked to a common resistance QTL (Figure 2). Additionally, we mapped the physical location of our significant markers with previously published markers associated with Sclerotinia stem rot resistance. With the exception of a novel position on chromosome 16, the majority of our markers confirmed previously identified genetic hot spots associated with Sclerotinia stem rot resistance (Figure 2).

After greenhouse evaluations, 31 lines including parental lines and the susceptible controls, Dwight and 91–44, were evaluated for Sclerotinia stem rot severity and the important agronomic traits of yield and lodging. Line performance in-field provided an assessment of the use and commercialization potential of lines. Significant differences among lines were observed in 2014 field tests for Sclerotinia stem rot severity (P < 0.0002) and incidence (P < 0.0001) (Table 5). The experimental line most susceptible to Sclerotinia stem rot based on DSI and DI in the 2014 field test was 91–44 which had a higher DSI and DI than all other lines. Lines SSR81–107, SSR51–70, and the parental lines had DSI and DI ratings significantly lower than 91–44 but not different from most lines (Table 5). AxN-1-55 yielded higher than all cultivars and lines, however, Dwight, 52–11, and 52–82B were not significantly different from this line. Therefore, all breeding lines were significantly lower in DSI and DI than the susceptible line, 91–44, and lines from population 52 had promising yields.

Similarly, lines were evaluated for Sclerotinia stem rot and agronomic traits in 2015. Differences between disease responses were also present in 2015 (P < 0.0001) (Table 5). In 2015, the susceptible check, Dwight, and line 91–44 had significantly greater DSI scores than any other line, 50.9 and 50.2, respectively. Lines SSR51–70, W04–1002, 52–82B, and 91–103 had among the lowest DSI scores (<5.0). Dwight, 52–14, AxN-1–55, and 52–82B were the highest yielding of all lines (>3,480 kg ha^-1), while the lowest yielding lines included the highly Sclerotinia stem rot-resistant lines W04–1002 and 91–145 (<2,300 kg ha^-1). Breeding lines again demonstrated better resistance than susceptible lines, and lines such as 52–82B exhibited both the desirable phenotypes of low disease and high yield.

Among the 10 lines evaluated in 2016, significant differences were present among lines for DSI (P < 0.01) and DI (P < 0.01) (Table 5). DSI and DI values were much higher in 2016 compared to previous years. The most susceptible lines were Dwight, 91–44, and 52–11 with respective DSI values of 91.8, 85.2, and 72.0. In addition to these lines, 52–82B also had a higher DI compared to all but three of the lines. Lowest DSI rankings occurred in lines SSR51–70 and W04–1002 with DSI values of 23.5 and 16.5, respectively. 91–38 also exhibited low disease severity levels that were not significantly different from SSR51-70. Yield in 2016 was greater than in previous years, and the highest yields occurred in lines AxN-1–55 and 52–82B, 3,865.2 and 3,822.9 kg ha^-1, respectively. Lines 91–145 and 52–11 yielded similarly to these lines. Lines with the lowest yields were Dwight and 91–44. Interestingly, results demonstrated high disease ranking for the high yielding varieties 52–11 and 52–82B and lower DSI for SSR51–70 compared to many lines, as previously observed. This outcome occurred despite unusually high disease levels in a field naturally inoculated with S. sclerotiorum infected sunflowers.

In 2016 field nurseries, 91–38 and AxN-1–55 had the least lodging (P < 0.05) with mean lodging scores that were not significantly different from high yielding lines, 52–11and 52–82B (Supplementary Table S2). Lines 91–145 and SSR51–70 exhibited the highest lodging scores, 3.4 and 3.2, respectively, and were not significantly different from 91 to 44, which had a score of 2.4. Differences in lodging ranks between lines were observed in 2014 (Supplementary Table S1) and 2016, but not 2015. 91–38 and 52–11 were not significantly different from AxN-1–55 in 2014 (Supplementary Table S1). Lodging results indicated the problematic trait of lodging was consistently present in the highly resistant line, SSR51–70. However, other lines such as 91–38 and 52–82B exhibited positive traits such as moderate disease resistance and high yield in conjunction with a good stand.

In 2015, protein and oil were added to the agronomic traits evaluated, as they are important considerations for commercialization and breeding (Supplementary Table S3). The selected breeding lines were also evaluated in 2016 (Supplementary Table S4). In 2016, the line with the highest protein content, 39.4%, was W04–1002 (P < 0.05), and it was not significantly different from SSR51–70 at 39.2 % and 91–44 at 38.7%. The highest percentage of oil, 19.2%, was measured from 91 to 44 and 91 to 38. Similarly, the aforementioned lines produced high levels of protein and oil in 2015 (Supplementary Table S2).

Overall, after 3 years of evaluations for disease responses and agronomic traits, genetic gain was observed within breeding populations. Desirable observed traits include high levels of disease resistance, as observed in SSR51–70, and maintained yields, as observed with 52–82B. Additionally, moderate disease resistance was observed in conjunction with high protein and oil or moderate yield as observed in 91–38 and 51–23, respectively. Therefore, field evaluations elucidated several promising lines for future breeding or commercialization.

To determine the physiological resistance of lines, in the absence of field escape mechanisms, greenhouse inoculations of the 2016 lines were conducted using nine previously characterized (Willbur et al., 2017) S. sclerotiorum isolates. Previously, 2013 greenhouse evaluations used only a single aggressive isolate. However, current S. sclerotiorum research indicates that various isolates may elicit differential resistance responses (Willbur et al., 2017). In 2016, differences in the AUDPC of lines, measured at 5, 10, and 14 DPI, were explained by the line inoculated (P < 0.01) and the isolate used (P < 0.01. The most resistant lines, as indicated by the lowest AUDPC, were 52–82B and 91–38 (P < 0.05) (Figure 3). AUDPC results were lower than the resistant parents,’ and these lines outperformed several lines considered more resistant based on results from field evaluations. For example, SSR51–70 was consistently more resistant in field experiments than 52–82B and 91–38. Genetic gain was once more observed, as 52–82B and 91–38 had lower AUDPCs than resistant parents. Additionally, 52–82B and 91–38 are promising lines for agronomic traits, as previously observed. Therefore, high levels of Sclerotinia stem rot resistance bolsters their applicability as commercial lines. Overall, 2016 multi-isolate greenhouse evaluations demonstrated the importance of pathogen diversity and screening in a controlled environment for physiological resistance and the broad applicability of breeding efforts.