Abstract
Most of the tomato (Solanum lycopersicum) varieties currently used in organic farming were bred for conventional farming, often characterized by high-input use. These varieties do not perform as well in low-input organic systems, generating the need to develop varieties that are adapted to organic management systems. This project focused on improving flavor, disease resistance, and yield, all identified as key traits by organic tomato farmers in the Upper Midwest, USA. Ten advanced tomato breeding lines and two check varieties were developed and evaluated for 16 traits in organic high tunnel systems in 2020 and 2021. The line CSDE-F6.47 averaged 6.32 kg/plant and obtained high flavor intensity and overall flavor scores (3.78 and 3.69 out of 5, respectively). The line JBDE-F5.31 was another outstanding line, with a yield of 5.18 kg/plant, with good flavor intensity (3.32) and overall flavor (2.92) scores. Broad-sense heritability of marketable weight per plant was high (0.91), and the genetic variance was also high, which shows the opportunity to continue to increase the marketable weight in lines with excellent flavor. A significant positive correlation was found between overall flavor and °Brix (0.56), and titratable acidity (0.70), indicating that both measurements can be good predictors of overall flavor. The most promising lines will be further evaluated on-farm to evaluate their potential as releasable varieties.
Tomato (Solanum lycopersicum L., 2n = 2x = 24) is an important crop worldwide. In 2019, 5.03 million hectares yielded 180 million tons of tomatoes (FAOSTAT 2019). In the United States, tomatoes accounted for 36% (10.85 million tons) of total vegetable production in 2019. These data include fresh market and processing tomatoes. Tomatoes are a locally important crop in the Upper Midwest of the United States. In 2014, the state of Wisconsin produced nearly 1.5 million pounds of tomatoes, and by 2021, the National Organic Program (USDA 2021a) reported that 259,616 pounds of tomatoes grown in open field (including for processing and fresh market) were certified organic. The fruits are rich in antioxidant molecules including carotenoids, ascorbic acid, vitamin E, and flavonoids (USDA 2021b). Interest in the health benefits of tomatoes has led to breeding projects focused on improving the nutritional quality of the fruit (Frusciante et al. 2007; Sacco et al. 2013; Sun et al. 2012).
Tomato is a key crop for diversified farmers in the Upper Midwest because of its high value compared with other crops. Farmers seek tomato varieties that perform well under local growing conditions while satisfying the needs of their market. Such varieties have been especially hard for organic farmers to find. According to the State of Organic Seed report (Hubbard and Zystro 2016), 82% of the respondent vegetable growers still relied on conventional seed for some part of their production system, with an average of 70% of their acreage under production using conventional seed. One of the reasons listed as to why they did not use organic seed was the lack of desirable traits available in an organic variety. It has been established that conventional breeding objectives can differ from organic breeding objectives and breeding for the specific needs of organic systems is essential to developing high-performing varieties for organic agriculture (Ceccarelli 1994; Lammerts van Bueren and Myers 2012). Environmental conditions on conventional farms are relatively similar to each other because of higher and more uniform use of synthetic fertilizers and pesticides, whereas more variable management practices in organic systems lead to field conditions that differ more significantly among organic farms. Because of the difference in the production conditions and the market objectives, the desirable traits for a variety differ for organic and conventional farmers.
Flavor is the top priority for tomato growers in the Midwest, followed by disease resistance, crack resistance, and nutritional value (Hoagland et al. 2015). In terms of disease concerns, 67% of the organic farmers responding to the survey found early blight (Alternaria solani) (EB) difficult to control, while 72% found Septoria leaf spot (S. lycopersici) (SLS), 58% found leaf mold (Passalora fulva) (LM), and 33% found late blight (Phytophthora infestans) (LB) hard to control. Powdery mildew (Oidium neolycopersici) (PM) was not identified as an important pest in this survey, but it has appeared to spread more recently in greenhouse and high tunnel tomato production, making it a potential disease of concern for current and future farmers. Genetic resistance to PM has been found to be related to six monogenic genes (Bai et al. 2005) and three polygenic Quantitative Trait Loci (QTLs) (Bai et al. 2003), and none of these are known to be present in the parental lines of the tomato breeding lines evaluated in this research. EB is a disease that is particularly important to organic farmers because there are few products that can help with prevention, and it is even more complicated to stop the spread once the fungus cycle has started. LM is a fungal disease that has become predominant in high tunnels, due to the high relative humidity and low ventilation that can potentially occur depending on the management (Sudermann et al. 2022). Combining disease resistance, high yields, and good flavor in tomato has proved challenging. SLS resistant tomato varieties have been developed, but they do not always have acceptable flavor. Although organic direct-market farmers report that flavor is the most critical trait that they consider when choosing varieties, they also require varieties that have sufficient production potential and disease resistance. According to a survey of Wisconsin organic vegetable farmers carried out in 2012 by Lyon et al. (2015), disease tolerance was the highest-ranked trait regarding priorities for plant breeding, followed by insect tolerance and yield.
Decentralizing the breeding process and involving farmers in trials can result in improved organic breeding outcomes (Casals et al. 2019; Dawson et al. 2011). Participatory plant breeding (PPB) enables farmers and breeders to develop varieties that are adapted to local conditions, and selection and trials can happen both in research stations and on-farm. PPB was initially used to support economically disadvantaged farmers in the Global South who were not benefitting from nonparticipatory, conventional breeding programs (Bellon 2006). A participatory approach can address several challenges related to conventional crop improvement. Decentralization of the environments where selection is carried out is key to developing varieties adapted to diverse agricultural systems. PPB itself promotes the diversification of environments and integration of multiple actors in the breeding process, working toward a more geographically and stakeholder decentralized variety development. Including farmers early in the breeding program can greatly accelerate relevant improvements, especially if they are experienced in the nuances of their production systems and market preferences. PPB approaches have been successful both in the Global North and South. Colley et al. (2021) identified 47 projects across the United States, Canada, and Europe, including trials with 22 crops. In the United States alone only, PPB has aided in the development of new varieties, genetic diversification, and conservation of staple crops such as apples, tomatoes, maize, oats, peppers, potatoes, and wheat, among others. A successful example of PPB is the development of ‘Who Gets Kissed’ sweet corn, in which organic farmers defined the priority traits for the breeding program, then researchers and a public sector plant breeder in Wisconsin carried out the initial population development. The project culminated with the release of a new variety that farmers in other states such as Oregon, Washington, California, and New Mexico have continued to adapt to their local conditions (Shelton and Tracy 2015).
Our project emerged as an initial step toward wider collaborative organic breeding efforts that can meet the overall demand for reliable organic tomato varieties while developing high-performing varieties that are specifically adapted for organic farming in the Upper Midwest. Previous work identified promising tomato varieties (Healy 2016; Hodge et al. 2019) that were chosen as parental varieties for our participatory breeding project. Organic farmers hosted production trials of our breeding lines; chefs evaluated their culinary qualities; and research trials assessed yield, production traits, and response to plant diseases. This article presents the results of the project and analyzes the potential of the breeding lines to be released as varieties or used as genetic resources for future tomato breeding efforts.
Materials and Methods
Plant material.
We evaluated 10 advanced breeding lines developed under organic management (Healy et al. 2017; Hodge et al. 2019). These lines were selected under an organic high-tunnel management system during the summer seasons, and advanced without selection in a greenhouse each winter. The parental lines were selected for their high quality, particular flavor, disease tolerance, and yield. Defiant is an F1 hybrid that has high resistance to LB (resistance genes Ph-2 and Ph-3) and intermediate resistance to EB (Johnny’s Selected Seed 2021). OSA404 is a cross between ‘Wisconsin 55’ (WI55) and a disease-resistant North Carolina State inbred, selected by the Organic Seed Alliance for disease tolerance and flavor over several years. It was received in 2014 as an advanced line and maintained by selfing. A6 is a reselection of an Amish heirloom selected for Midwestern adaptation by Craig Grau, a retired plant pathologist at University of Wisconsin–Madison. ‘Japanese Black Trifele’ is an heirloom with a smoky flavor maintained by K Greene at the Hudson Valley Seed Company. ‘Crimson Sprinter’ is an heirloom from Ontario, CA, USA, with partial SLS resistance, earliness, and good flavor. ‘Defiant’ and ‘Japanese Black Trifele’ were also used as check varieties. A summary of the trialed breeding lines can be found in Table 1. Figure 1 shows a simplified version of the crossing diagram used to develop the 10 breeding lines evaluated in this project.
Parent lines, generation, market type, and fruit color and size of 10 tomato breeding lines evaluated in 2020 and 2021 in organic high tunnel systems in the West Madison Agricultural Research Station, Verona, WI, USA.
Summarized crossing diagram used to develop 10 tomato breeding lines in Madison, WI, USA.
Citation: HortScience 59, 9; 10.21273/HORTSCI17799-24
Line selection and advancement.
In Summer 2017, 18 tomato varieties, including commercial lines and unreleased advanced germplasm, were planted in both the open field and high tunnel, and 22 crosses were made between the parental lines. Of the 22 crosses, 16 were successful. In Winter 2017–18, the first F1s where planted, along with the parental lines for the crosses that were not successful in the first round. Three plants per F1 cross were planted in winter, and seed was saved from at least one plant. The six remaining crosses were successfully made during that winter as well. In Summer 2018, the F1s obtained from the winter crosses and the F2s obtained from the seed advancement were planted in a high tunnel at the West Madison Agricultural Research Station (WMARS), totaling 22 families and 40 lines in total. The experimental unit consisted of three plants per breeding line, and each was replicated twice. One or two plants were selected per cross for advancement, taking into consideration productivity, flavor, and disease resistance. At this point, 55% of the lines were dropped from the program, keeping eight families and 18 lines total. The seed saved in Summer 2018 was advanced during Winter 2018–19, obtaining F3 and F4 seed. The F3 and F4 lines were planted in Summer 2019, totaling 18 lines. From Summer 2017, two families were dropped from the program, keeping six families and 10 lines in total. The 10 lines were advanced during Winter 2019–20, and the selected families and lines were planted in Summer 2020 and 2021 following the experimental design explained in the next section. Figure 2 summarizes the selection and advancement process followed in this tomato breeding program.
Selection and seed advancement diagram used to develop 10 advanced tomato breeding lines in Madison, WI, USA from 2017 to 2021.
Citation: HortScience 59, 9; 10.21273/HORTSCI17799-24
Experimental design.
In 2020 and 2021, the advanced breeding lines were grown at the West Madison Agricultural Research Station, University of Wisconsin–Madison, Madison, WI, USA (43.06054765°N, 89.52376954°W, elevation 323 m) on land certified organic by the Midwest Organic Services Association since 2008. The high tunnel was on a level area, oriented north-south; the rows ran from east to west, with the long ends facing east and west. The high tunnel dimensions were 30 ft × 88 ft (9.1 m × 26.8 m) and covered ∼2640 ft2 (245.3 m2). The experiment was designed as a randomized complete block design. The experimental unit (plot) comprised three individual plants in 2020 and four individual plants in 2021. The experimental units were replicated twice in both years. The within-row plant spacing was 2 ft (0.61 m), and beds were 4 ft apart (1.2 m), center to center, with 2-ft (0.61 m) aisles.
Management.
In 2020 and 2021, the soil was amended according to soil nutrient analysis using Renaissance 11–0–0, an Organic Materials Review Institute (OMRI) approved feather meal fertilizer (PJC & Company, Ecological Land Care Inc., Rowley, MA, USA) to achieve a total rate of 125 lb N per acre (140 kg/ha). Before each planting, a cover crop of winter rye was grown and incorporated. In 2020, transplants were started by West Star Organics in Cottage Grove, WI, USA, a USDA-certified organic grower. In 2021, the transplants were started by Circadian Organics in Viroqua, WI, USA, also a USDA-certified organic grower. Transplants were planted on 12 May in both years. Beds were laid with drip irrigation and covered with black landscape fabric. The aisles were covered with straw mulch to control weeds. Plants were watered consistently 3 d a week during the first month, decreasing to twice a week from then on. Plants received between 0.75 and 1.0 inches of water per week.
Pruning was done weekly until the plants were ∼5 ft tall. After this, pruning was done as required to maintain a two-lead trellis system. When the plants reached 4 to 5 ft tall, the basal leaves were pruned to increase the airflow of the canopy and to increase the distance between the leaves and the soil, with the purpose of diminishing the potential of disease development and spread. The plants were trellised using tomato clips to guide the plants up the twine attached to the cross-braces. The high tunnel temperature was closely monitored to decide when to open and close the sides. Side vents and doors were opened manually to maintain good ventilation and keep day-temperatures below 95 °F (35 °C) and above 65 °F (18 °C) during night. Higher than 95 °F daily temperatures can cause detrimental effects on pollen quality, pollination, flower abortion, and fruit set.
Harvest.
Harvest was done once a week, picking tomatoes that were at least in “pink stage” following the USDA definition (USDA 2005), which means that at least 30% of the fruit surface, in aggregate, shows a change in color from green to tannish yellow, pink, or red. Yield was recorded on a per-plot basis. At harvest, the tomatoes were subdivided into marketable and unmarketable fruit. Marketable fruit was counted and weighed, whereas unmarketable tomatoes were weighed and the reason(s) of unmarketability recorded. Weights are presented in kilogram/plant and tons/plant, where ton refers to metric ton throughout this article. Any fruit with signs of damage such as splitting, cat face, windowing, scarring, insect, rodent signs, and rots were considered unmarketable. A sample of fully ripe fruit was set aside for the quality evaluation. The percentage of unmarketable fruit by weight was also calculated for each plot.
Fruit flavor.
Tomato fruit flavor was evaluated weekly. In 2020, samples were packed individually for research staff to take home. This was implemented to follow the sanitary guidelines in effect due to the coronavirus pandemic. In 2021, research staff conducted in-person tasting in the field three times during the peak harvest season. The tasting group participated in a calibration exercise at the beginning of the season. This exercise included the recognition of the basic flavor components—sweet, acid, salty, bitter, and umami—at three different concentrations in water. Breeding lines were divided into different tasting groups based on shared parents and market similarities. When a breeding line did not have enough fruit for testing in its designated group and date, it was tasted later in the season. Only completely ripe fruit was used for flavor analysis. Fruit from each plot of each line was bulked in a composite sample. Tasters rated each sample on a 1 to 5 scale for sweetness, acidity, saltiness, bitterness, and umami, where 1 was very low perception and 5 was high perception of that flavor. Flavor intensity was also rated on a 1 to 5 scale, with 1 being low, and 5 being high intensity of a “tomato” flavor. Finally, after completing the tasting set, tasters were asked to return to each sample and rate it on an overall scale for flavor with 1 being very bad and 5 being excellent.
Because the taste testing includes the participation of people, an institutional review board (IRB) application was submitted to the University of Wisconsin–Madison for approval. The IRB committee qualified the study to be exempt from federal regulations under category 45CFR 46.101(b)(6): Taste and food quality evaluation (Pre-2018 Requirements), and the protocol was approved on 26 Aug 2014 and renewed every 3 years.
Fruit citric acid equivalents and °Brix.
A sample of the fruit set apart for tasting was saved and frozen to later measure citric acid equivalents (CA) and °Brix (as a measure of total dissolved solids) in the laboratory. This was to analyze and find any correlations between the taster’s flavor perception and some of the main elements, sugar, and acid, which contribute to tomato flavor. For both CA and °Brix, tomato juice samples were filtered through cheese cloth to remove excess solids. °Brix levels were tested using a digital refractometer (model no. 30051; Sper Scientific, Scottsdale, AZ, USA). Acidity content by volume was measured using an automatic titrator (model Hi901C; Hanna Instruments, Smithfield, RI, USA) and is reported as g citric acid/100 g sample because CA is the main organic acid in tomatoes that contributes to titratable acidity and pH (Wilkerson et al. 2013). Twenty milliliters of each tomato sample were titrated, NaOH was used at a 0.1 N concentration, and the samples were titrated to a fixed endpoint of 8.1 pH.
Disease incidence and severity.
Plots were scored every 2 weeks starting 8 weeks after transplanting. Diseased foliage was recorded using a 0% to 100% scale in 5% increments of leaf area affected by disease. A 0% score referred to a healthy plot that had no symptoms of foliar diseases. A 100% score referred to a plot where the plants were completely dead. The evaluation included EB, LM PM, and SLS. The area under the disease progress curve (AUDPC) was calculated for each breeding line. The AUDPC is a quantitative summary of the disease intensity over time and is useful to compare values across plant varieties, years, locations, or managements. The trapezoidal method was followed to calculate the AUDPC values for each disease for each breeding line by calculating the average disease intensity between each pair of adjacent data points (weekly) (Jeger 2004).
Statistical analysis.
where
Correlations.
The production, fruit quality, and disease traits were correlated to each other by calculating the Pearson correlation coefficient. The correlation matrix was calculated using the cor base function in R Programming Software. A year-to-year Pearson correlation coefficient was also calculated for each trait using the cor function in R. Correlations reported are Pearson coefficients (
Broad-sense heritability.
where
where VG is the genotypic variance, VGE is the genotype-by-year variance, Vɛ is the error variance, r is the number of replicates, e is the number of years, MSError is the error mean square, MSGE is the genotype-by-year mean square, and MSGenotype is the genotype mean square.
On-farm trials.
The breeding lines evaluated in this project were trialed on-farm by six organic farmers in the Upper Midwest in 2020 and 2021. Farmers were sent enough seed to plant six plants per breeding line on their farm, following their normal agricultural management. The six plants were planted in a single plot, and each trial was planted within their normal production tomato area following the planting map that was provided (Supplemental Fig. 1), surrounded by other tomato varieties that farmers grew commercially. This allowed farmers to compare the breeding lines to the main tomato cultivars they would grow for market. The farmers were requested to submit their evaluations considering their own market needs (Supplemental Table 1).
The farmers grew the breeding lines under protected structures such as high tunnels or caterpillar tunnels. Throughout the season, the farmers took notes on overall yield, fruit quality, disease resistance, and earliness of each breeding line. Each farmer returned an evaluation form with notes on each line, and this feedback was used to understand the uniformity of the lines among different growing environments and to know how well these lines satisfied the needs of the farmers and the potential for variety release.
Results
Production
Genotype was a highly significant source of variation for all the production traits, which includes marketable count, marketable weight, proportion unmarketable by weight, and average fruit weight (Table 2). The broad-sense heritability was high for the production traits (Table 2). Year and genotype-by-year interaction were not significant sources of variation for these traits. Overall, the average marketable weight of all the breeding lines was 6.31 kg/plant and 6.11 kg/plant for 2020 and 2021, respectively, with no significant differences between years. No genotype-by-year interaction was observed for any of the production traits (Table 2).
Broad-sense heritability (H2), significance of F tests of production, disease, and fruit quality traits and year-to-year correlation evaluated from 10 tomato breeding lines grown in an organic high tunnel system in 2020 and 2021, at the West Madison Agricultural Research Station, Verona, WI, USA.
In 2020, the check varieties ‘Defiant’ (8.64 kg/plant) and ‘Japanese Black Trifele’ (8.12 kg/plant) had the highest marketable weight values, followed by the breeding lines CSDE-F6.46 (7.39 kg/plant), O4DE-F5.44 (7.06 kg/plant), and O4DE-F5.43 (7.04 kg/plant), with no significant differences between any of these lines (Supplemental Table 2). In 2021, the lines with the highest marketable weight were O4DE-F5.44 (8.29 kg/plant), ‘Japanese Black Trifele’ (7.59 kg/plant), O4DE-F5.43 (7.54 kg/plant), ‘Defiant’ (7.21 kg/plant), and CSDE-F6.46 (6.61 kg/plant), with no significant pairwise differences. In terms of fruit size, O4DE-F5.43 had the highest fruit weight (398 g/fruit), followed by O4DE-F5.44 (387 g/fruit), with no significant difference between these lines. These varieties can be considered large-sized slicers. The CSDE-F6.46 (135 g/fruit) and CSDE-F6.47 (135 g/fruit) lines were not significantly different from the check varieties ‘Defiant’ (112 g/fruit) and ‘Japanese Black Trifele’ (152 g/fruit) (Supplemental Table 2). In terms of proportion unmarketable, CSDE-F6.46, JBDE-F5.31, and ‘Defiant’ had the lowest values (0.10, 0.11, and 0.12 respectively). The lines O4JB-F5-MV1.116 (0.48), A6JB-F5.34 (0.42), and O4A6-F4-MV1.109 (0.41) had the highest proportion unmarketable.
Disease
Genotype and year were both significant sources of variation for EB AUDPC. No significant genotype-by-year interaction was observed. In 2020 and 2021, JBDE-F5.28, ‘Defiant’, and ‘Japanese Black Trifele’ had the lowest EB AUDPC scores, but differed significantly only from O4JB-F5-MV1.116 with the highest score (637). Year was a significant source of variation, and overall, the AUDPC scores for EB were higher in 2020 than in 2021. The year 2020 had a higher number of intense rain and elevated temperature events than 2021. EB thrives on wet surfaces (leaves) and moderate temperatures (∼27 °C) (Foolad et al. 2008); thus, these results align with expectations. The breeding line JBDE-F5.28 had the lowest EB AUDPC score (125), and O4JB-F5-MV.116 had the highest score (637) (Fig. 3). The broad-sense heritability of EB AUDPC was medium (0.67), with a high genetic variance. This can be explained by the high variability between the parental lines.
Marketable weight (kilogram/plant) (x axis), EB AUDPC (y axis), and °Brix (color graduation and circle size of the points) least squares means of 10 tomato breeding lines and two checks grown in an organic high tunnel system in 2020 and 2021, at the West Madison Agricultural Research Station, Verona, WI, USA.
Citation: HortScience 59, 9; 10.21273/HORTSCI17799-24
For SLS AUDPC, only year was a significant source of variation. In 2020, the overall SLS AUDPC score was 55, whereas in 2021, it was 398. Overall, CSDE-F6.47 had the highest score (459), and ‘Japanese Black Trifele’ had the lowest (53). Although there was not a statistically significant effect of genotype, the correlation among breeding line scores between years was moderate at 0.37, and broad-sense heritability was also moderately high (0.67), suggesting that with more consistent disease pressure, we might see more consistent differences among varieties.
For LM AUDPC, genotype was the only significant source of variation (Table 2), although there were not any significant pairwise differences. The AUDPC score ranged from 311 to 878 in 2020 and from 359 to 1379 in 2021. Overall, LM spread later in the season and did not seem to significantly affect yield or fruit quality. LM had a medium to high broad-sense heritability (0.71). This can be attributed in part to the variability between the parental lines.
For PM AUDPC, genotype, year, and the interaction between genotype and year were significant sources of variation. There was a significant difference between years, in the opposite direction of the SLS disease pressure. For PM AUDPC, the mean for 2020 was 838, and dropped to 140 in 2021. Overall, O4DE-F5.44 had the lowest PM AUDPC score (322), and CSDE-F6.47 had the highest (812). The broad-sense heritability for PM AUDPC was exceptionally low (0.12), which can be explained due to having a large proportion of the phenotypic variation explained by the genotype-by-year interaction (Table 2).
Fruit quality
Genotype was a significant source of variation for °Brix and CA. Year and genotype-by-year interaction were not significant sources of variation for either trait (Table 2). Genotype was a significant source of variation for all the taste tasting traits evaluated. For appearance, O4JB-F5.5 had the highest score (4.2); for texture, CSDE-F6.47 and CSDE-F6.46 had the highest scores (3.8); for sweetness, CSDE-F6.47 had the highest score (3.4); for acidity, CSDE-F6.47 had the highest score (3.4); for bitterness A6JB-F5.34, CSDE-F6.46, and O4JB-F6.35 had the lowest score (1.7); for umami, CSDE-F6.47 had the highest score (3.2); for flavor intensity, CSDE-F6.47 had the highest score (3.8); and for overall flavor, CSDE-F6.47 had the highest score (3.7) (Supplemental Table 2). The broad-sense heritability of the tasting attributes was low in general (Table 1). This illustrates the challenge of selecting for flavor because it is hard to obtain enough data from taste evaluations to have sufficiently high heritability for selection. However, tasting evaluations provide information that is not available in other ways. In this case, combining information from tasting evaluations with indirect selection based on correlated laboratory measurements might be a more efficient approach. The correlation among overall flavor, flavor intensity, and °Brix is high (Fig. 4), and the broad-sense heritability of °Brix is moderately high, suggesting that indirect selection may help increase genetic gain for flavor when selecting on multiple traits.
Correlation plot between production and fruit quality traits evaluated in 10 tomato breeding lines organically grown in high tunnels in 2020 and 2021 at West Madison Agricultural Research Station, Verona, WI, USA.
Citation: HortScience 59, 9; 10.21273/HORTSCI17799-24
Correlations
Marketable weight was negatively correlated with °Brix (−0.45), CA (−0.44), acidity (−0.31), and flavor intensity (−0.27). °Brix is positively correlated with sweetness (0.43) and flavor intensity (0.73). CA is positively correlated with acidity (0.79) and flavor intensity (0.86) (Fig. 4). The overall flavor rating was highly correlated to °Brix (0.57), CA (0.69), sweetness (0.85), umami (0.65), and flavor intensity (0.92) (Fig. 4). Average fruit weight also had a negative correlation with °Brix (−0.68) and CA (−0.85). This means that larger fruit had lower °Brix and CA. Something similar can be observed in the correlations between average fruit weight and flavor intensity (−0.72), or overall flavor ratings (−0.51). These negative correlations may suggest the trade-offs between size and fruit quality; improving fruit quality could mean reducing the fruit size if fruit size is not controlled during selection. In terms of year-to-year correlations, all the production traits had high significant values (>0.8, Table 2). EB AUDPC had a medium year-to-year correlation (0.66). Of the fruit quality traits, CA (%), appearance, and overall flavor had high significant year-to-year correlation coefficients.
Results by family
‘Crimson Sprinter’ by ‘Defiant’ (CSDE) Family.
CSDE-F6.46 and CSDE-F6.47 both had high marketable weight (7.00 and 6.32 kg/plant, respectively), with no significant differences with each other or when compared with ‘Defiant’ (7.87 kg/plant). Both lines had some of the lowest proportion unmarketable of all the breeding and check lines (Supplemental Table 2). The lines were not significantly different for EB AUDPC scores compared with ‘Defiant’. High LM AUDPC scores were observed in 2021 (Supplemental Table 3) but were not significantly different from any of the other check varieties or breeding lines. The spread of LM was late in the season, with overall incidence higher than 40% starting week 35 in 2021 (23 Aug). Line CSDE-F6.47 had the highest flavor intensity (3.8) and overall flavor (3.7) scores. Line CSDE-F6.47 was not significantly different in yield from its sister line CSDE-F6.46 and had significantly higher scores for flavor intensity and overall flavor. In this case, improving flavor did not translate into a lower marketable weight, as was expected with the obtained negative correlation between production and fruit quality traits. The farmers shared similar observations. They considered this family to have a good acid/sweet balance, and for some, it was the best tasting family of all. Fruit size on-farm was medium, as it was at on-station trials, with an average fruit weight of 135 g (Supplemental Table 2). The line CSDE-F6.46 averaged 69.5 ton/ha, making it a strong candidate for potential release. A comparable commercial variety is ‘Defiant’, one of the parental lines for these breeding lines. In 2016, ‘Defiant’ had a yield of 69.42 ton/ha in 2016 (Seed to Kitchen 2016), 72.00 ton/ha in 2020, and 59.17 ton/ha in 2021.
OSA404 by ‘Japanese Black Trifele’ (O4JB) Family.
The two breeding lines of this family were significantly different from each other for most traits. O4JB-F6-5 performed significantly better for most traits, so the following discussion focuses exclusively on this line. O4JB-F6-5 had high marketable weight (6.13 kg/plant) and was not significantly different from ‘Defiant’ or the CSDE family lines. O4JB-F6-5 had a low proportion unmarketable. This line was highly rated in the tastings. In appearance it got the highest score. The flavor intensity score was 3.1, while the overall flavor score was 2.9, both falling into the median range. There were no significant differences in EB AUDPC scores compared with ‘Defiant’ and the CSDE family lines. The same was observed for LM AUDPC. Farmers liked the size and the flavor of this breeding line, which was similar to the heirloom parents. For some, this was the best breeding line for production traits, with sustained good flavor (Table 3). ‘Genuwine’ is a comparable commercial variety. It produces a large beefsteak-type tomato with an average fruit weight of 265 g/fruit, and has broad disease resistance, although based on ratings from our trials it does not have good flavor intensity (Seed to Kitchen 2016). ‘Paul Robeson’ and ‘Pruden’s Purple’ are two heirloom-type varieties that farmers like in the Upper Midwest, due to their consistently highly preferred flavor, although some find ‘Pruden’s Purple’ to be too large. Previous trials at the same research station show ‘Paul Robeson’ with a marketable weight of 33.25 ton/ha (Seed to Kitchen 2016), and ‘Pruden’s Purple’ at 43.9 ton/ha (Seed to Kitchen 2015). O4JB-F6-5 had significantly higher marketable yield (60.9 ton/ha) with similar fruit quality characteristics, making it a breeding line with high potential for commercial release, especially for local gardeners and diversified organic farmers.
Summary of on-farm evaluations of tomato breeding lines grown by organic farmers in the Upper Midwest United States in 2020 and 2021.
‘Japanese Black Trifele’ by ‘Defiant’ (JBDE) Family.
The breeding lines JBDE-F5.28 and JBDE-F5.31 were not significantly different from each other for any of the production traits. The average marketable weight was 5.19 and 5.18 kg/plant, respectively, lower than the CSDE lines and ‘Defiant’. These lines had a very low proportion of unmarketable fruit (<0.14) (Supplemental Table 3). These lines performed similarly to the CSDE family lines in terms of EB AUDPC and LM AUDPC scores. Flavor intensity scores were above 3.0, and overall flavor was 3.0 and 2.9, respectively. Farmers liked the flavor but noted that the fruit of these lines had a tougher skin than other breeding lines. Productivity varied from farm to farm (Table 3).
A6 by ‘Japanese Black Trifele’ (A6JB) Family.
The breeding line A6JB-F5-34 was characterized for producing large, pink-colored tomatoes. Coming from a cross between two heirloom varieties, it can be classified as a contemporary heirloom. Its yield was not the highest (4.55 kg/plant), but the general-public tasters liked the fruit because of its color, size, and intense flavor.
Discussion
The high broad-sense heritability values observed for the productivity traits and some of the diseases in this study have important implications for future breeding programs. The production traits marketable weight, marketable count, and average fruit weight all had high broad-sense heritability. This indicates that in our organic high tunnel system, we are seeing consistent variety rankings and low levels of genotype-by-environment interactions. However, it is important to consider the limitations of these heritability values. The study evaluated the lines in only one location over a period of 2 years. This limited scope may lead to inflated heritability values compared with what might be observed in a broader range of environments. Additionally, the wide range of genetic backgrounds in the trial may have contributed to increased genetic variance and broad-sense heritability. The lines evaluated are advanced enough that most alleles are close to being fixed, which decreases the dominance variance, leaving the additive variance to account for the majority of the genetic variance.
Two areas that could help further understand genotype-by-environment interactions are the use of multiple research station locations and comparison with a different growing system, such as open-field cultivation. This project was initially intended to be trialed in two locations, but because of the COVID-19 pandemic, work and travel restrictions in 2020 and 2021 did not allow for a second trial location to happen. Astroza (2020) evaluated six commercial lines including ‘Defiant’ and ‘Japanese Black Trifele’, at the West Madison and Spooner Agricultural Research Stations in Wisconsin in a 2-year trial. From the six lines, there was only one genotype with crossover interaction between the locations for the production traits, with no genotype-by-environment interaction effect for °Brix. Most farmers in the region are shifting from open-field tomato cultivation to the use of a protected structure (i.e., caterpillar tunnel or high tunnel) for their highest value tomatoes, including heirlooms and others with good flavor, and focus on canning or paste varieties with strong disease resistance for the open field. Given demand for varieties that are adapted to organic high tunnel systems, we prioritized the evaluation of the tomato lines under high tunnel production. We also prioritized trials at on-farm sites to have a better representation of the diversity of organic growing systems, which is difficult to replicate in on-station trials. Although the gathered data can provide valuable information about the genetic control of traits, it is essential to continue to validate these findings in multiple locations and growing seasons to ensure their reliability.
Disease resistance is a high priority, both for outdoor and high tunnel production. Although critical in the field, EB and SLS do not appear to be of concern in high tunnels due to the physical protection the tunnel provides, limiting free water on leaves and the potential spread of spores (Foolad et al. 2008). The advanced tomato breeding lines evaluated here showed low to medium EB and SLS infection, without a significant decrease in yield. The advanced lines evaluated in this project showed a medium level of tolerance to LM and PM, and the onset and spread of the disease was late enough in the season that it can be concluded it did not affect overall marketable weight. Good agricultural management, such as opening the sides of the tunnel in the mornings to promote air flow, can slow down the spread of the pathogen. Interestingly, the heritability of PM AUDPC (0.12) was lower than the other diseases. This could be because there was little genetic variation for resistance; none of the parental lines, and therefore the breeding lines, had a known resistance gene or QTL that could confer significant differences in the disease scoring between breeding lines. Additionally, the correlation of breeding line scores between years for PM AUDPC was negative, which may be a result of the significant variability in PM pressure between years and genotype-by-year interactions observed for the lines evaluated. The large variance due to genotype-by-environment interactions negatively affects the calculated heritability.
For quality traits, although heritability values were low, we found strong correlations that may help simplify evaluation, while still allowing selection on quality in early generations. The correlation between overall flavor and flavor intensity was 0.91 (P < 0.001). Baldwin et al. (1998) found equivalent results, obtaining an 0.71 correlation between the same traits. This suggests that consumers prefer more intensely flavored fruit. Merscher (2020) also found a significant correlation between overall flavor and flavor intensity (0.74) and overall flavor and sweetness (0.7) when evaluating pink and red slicer tomato varieties. In parallel, CA was highly correlated with acidity (0.82), flavor intensity (0.87), and overall flavor (0.70), and °Brix was highly correlated with flavor intensity (0.72) and overall flavor (0.56). This suggests that °Brix and CA could be good measures to predict flavor. An analysis of specific sugars, such as glucose and sucrose, could potentially give better estimates of flavor. It is important to note that the perception of flavor was also highly correlated to the °Brix/CA ratio because the acid content plays an important role in the taste of sugar during the consumption of tomato fruits (Wang et al. 2023). The correlation found between °Brix and CA with overall flavor acceptance is high enough in this analysis to warrant its future use for flavor predictability calculations alongside sensory evaluations. In addition, the high correlation between flavor intensity and overall liking may allow for a simplified taste test during early generations, allowing researchers to evaluate more breeding lines while still conducting a more extensive taste evaluation on advanced lines such as we present in this article. Similar to what Merscher (2020) found, the use of tasting analysis offered flexibility to the researchers, allowing us to answer simple questions about the flavor components of the tomato lines, rather than carrying out time-consuming volatile compound analysis. Interestingly, the overall flavor perception and texture had a high year-to-year correlation, whereas the individual flavor components did not (Table 2).
In general, flavor and marketable weight were negatively correlated. The correlation of marketable weight with CA was −0.45 and with °Brix was −0.44. Prior research using whole-genome sequencing and genome-wide association analysis has allowed researchers to associate the loss of high-sugar alleles with domestication and improvement, as larger fruits were selected (Tieman et al. 2017). This means that the selection for larger fruits produced the loss of specific alleles that contribute to higher °Brix content, making current selection for both higher fruit weights and higher soluble solids content difficult. De Souza et al. (2012) evaluated the yield, °Brix, and CA of the F1s obtained from a diallelic cross between five parental lines. They obtained a correlation of −0.18 between plant yield and °Brix and −0.13 between plant yield and CA. Tieman et al. (2012) found that aroma volatiles make important contributions to the perceived sweetness of tomato fruits. By reducing unpleasant volatile compounds, which may have simpler genetic inheritance than perceived sweetness, breeders could more easily improve consumer perception of flavor (Zhao et al. 2019). However, this would require the development of molecular markers for the relevant volatile compounds in appropriate genetic backgrounds.
Selecting only for higher marketable weight without evaluating flavor could negatively affect the overall flavor perception by consumers, confirming Klee and Tieman’s (2013) findings. This can make the future selection process difficult because the main objectives of this breeding program are to generate varieties with high yields and excellent flavor, adapted to organic high tunnel systems. For this reason, it is important to evaluate flavor early in selection, even if heritability is lower, to avoid eliminating breeding lines with good flavor. Working with indirect measures such as °Brix and CA in addition to simplified flavor evaluation may also help maintain excellent flavor in early generations of selection while also selecting for production and disease resistance.
From verbal communication with the tasters, the breeding lines had good flavor overall, and overall flavor scores were higher than for the check varieties in the trial. We received similar feedback from the farmers who tried the breeding lines in their production systems: overall flavor was good, and flavor was outstanding in some of the breeding lines. Before the COVID pandemic, the breeding program did regular evaluation of breeding lines with local chefs. This was interrupted in 2020 and 2021 due to pandemic restrictions. We resumed in a limited way in 2022 and 2023 and received feedback from chefs on overall flavor of the advanced lines, similar to farmer evaluations of the lines. This was done after the completion of the study reported here but served to confirm the selections we had made based on chef involvement before the pandemic and our research team flavor evaluations. This demonstrates that this type of selection program is a promising starting point for flavor improvement while maintaining adequate production and disease tolerance.
The participatory aspect to this research was key to achieve well-adapted tomato varieties. Both the researchers and the participating organic farmers defined the objectives of this breeding program. Even though a participatory breeding approach is not exclusive to an organic breeding program, it is rare to find a conventional breeding program that involves farmer selection. The selection and evaluation of the breeding lines in organic environments on-station and on-farm make this tomato-breeding program different from a typical conventional breeding program. Evaluation and selection were conducted in and for organic farming systems. In this project, we prioritized fruit quality and adaptation to the specific conditions of organic high tunnels in the Upper Midwest. Trials conducted at research stations may not always be the most representative of organic production environments, even when certified organic. For this reason, we also prioritized having more on-farm evaluations over a higher number of on-station locations, providing us with farmers’ input that was key to carry out the selection process.
With the data collected on-station, on-farm, and from the taste tastings, there are four breeding lines that show high potential for commercial release: CSDE-F6.47, O4JB-F5.5, JBDE-F5.28, and O5DE-F5.43. These lines are being further evaluated on-station and on-farm, and depending on the additional data obtained, some or all of them will continue the process to be released as varieties to have available for farmers.
Conclusions
Overcoming the negative correlation between yield and fruit quality traits such as °Brix, CA, sweetness, flavor intensity, and overall flavor persists as a breeding challenge in the improvement of tomato varieties. The use of good tasting heirloom varieties as parental lines showed positive results in obtaining breeding lines with a favorable combination of yield and flavor, such as CSDE-F6.46 and CSDE-F6.47. Tomato breeding lines adapted to organic farming systems are key to organic farmers achieving the production goals and fruit quality standards that they need for their markets. The best breeding lines evaluated in this experiment will be further evaluated on-station and on-farm, and different options will be explored for their future commercialization. These lines are uniform enough that they can also be used as parent lines in future efforts, considering their improved fruit quality and sustained moderate to high marketable weight.
References Cited
Astroza J. 2020. Improving farmer options for sustainable and profitable direct-market tomato production with season extension in the Upper Midwest of the United States (MS Thesis). University of Wisconsin–Madison.
Bai Y, Huang C-C, Van Der Hulst R, Meijer-Dekens F, Bonnema G, Lindhout P. 2003. QTLs for tomato powdery mildew resistance (Oidium lycopersici) in Lycopersicon parviflorum g1.1601 co-localize with two qualitative powdery mildew resistance genes. Mol Plant Microbe Interact. 16(2):169–176. https://doi.org/10.1094/MPMI.2003.16.2.169.
Bai Y, Van Der Hulst R, Bonnema G, Marcel TC, Meijer-Dekens F, Niks RE, Lindhout P. 2005. Tomato defense to Oldium neolycopersici : Dominant OI genes confer isolate-dependent resistance via a different mechanism than Recessive oI-2. MPMI. 18(4):354–362. https://doi.org/10.1094/MPMI-18-0354.
Baldwin EA, Scott JW, Einstein MA, Malundo TMM, Carr BT, Shewfelt RL, Tandon KS. 1998. Relationship between sensory and instrumental analysis for tomato flavor. J Am Soc Hortic Sci. 123(5):906–915. https://doi.org/10.21273/JASHS.123.5.906.
Bates D, Mächler M, Bolker BM, Walker SC. 2015. Fitting linear mixed-effects models using lme4. J Stat Softw. 67(1):1–48. https://doi.org/10.18637/jss.v067.i01.
Bellon MR. 2006. Crop research to benefit poor farmers in marginal areas of the developing world: A review of technical challenges and tools. CAB Rev: Perspect Agric Vet Sci Nutr Nat Resour. 1(70). https://doi.org/10.1079/PAVSNNR20061070.
Bernardo R. 2020. Breeding for quantitative traits in plants. Stemma Press, Woodbury, MN, USA.
Casals J, Rull A, Segarra J, Schober P, Simó J. 2019. Participatory plant breeding and the evolution of landraces: A case study in the organic farms of the Collserola Natural Park. Agronomy. 9(9):486. https://doi.org/10.3390/agronomy9090486.
Ceccarelli S. 1994. Specific adaptation and breeding for marginal conditions. Euphytica. 77(3):205–219. https://doi.org/10.1007/BF02262633.
Colley MR, Dawson JC, McCluskey C, Myers JR, Tracy WF, Bueren E V 2021. Exploring the emergence of participatory plant breeding in countries of the Global North—A review. J Agric Sci. 159(5-6):320–338. https://doi.org/10.1017/S0021859621000782.
Dawson JC, Rivière P, Berthellot JF, Mercier F, de Kochko P, Galic N, Pin S, Serpolay E, Thomas M, Giuliano S, Goldringer I. 2011. Collaborative plant breeding for organic agricultural systems in developed countries. Sustainability. 3(8):1206–1223. https://doi.org/10.3390/su3081206.
FAOSTAT. 2019. Worldwide tomato production. http://www.fao.org/faostat/en/#data/QC. [accessed 14 May 2021].
Foolad MR, Merk HL, Ashrafi H. 2008. Genetics, genomics and breeding of late blight and early blight resistance in tomato. CRC Crit Rev Plant Sci. 27(2):75–107. https://doi.org/10.1080/07352680802147353.
Frusciante L, Carli P, Ercolano MR, Pernice R, Di Matteo A, Fogliano V, Pellegrini N. 2007. Antioxidant nutritional quality of tomato Research Article Antioxidant nutritional quality of tomato. Molecular Nutrition Food Res. 51(5):609–617. https://doi.org/10.1002/mnfr.200600158.
Healy GK. 2016. Tomato variety trials and selection for local adaptation and culinary quality in organic systems (MS Thesis). University of Wisconsin–Madison.
Healy GK, Emerson BJ, Dawson JC. 2017. Tomato variety trials for productivity and quality in organic hoop house versus open field management. Renew Agric Food Syst. 32(6):562–572. https://doi.org/10.1017/S174217051600048X.
Hoagland L, Navazio J, Zystro J, Kaplan I, Vargas JG, Gibson K. 2015. Key traits and promising germplasm for an organic participatory tomato breeding program in the US Midwest. HortScience. 50(9):1301–1308. https://doi.org/10.21273/HORTSCI.50.9.1301.
Hodge T, Healy K, Emerson B, Dawson J. 2019. Comparing tomato varieties under organic high tunnel and open field management in the north central region. https://eorganic.org/node/28426. [accessed 11 Jul 2023].
Hubbard K, Zystro J. 2016. State of organic seed, 2016. Organic Seed Alliance. https://seedalliance.org/all-publications/. [accessed 11 Jul 2023].
Jeger MJ. 2004. Analysis of disease progress as a basis for evaluating disease management practices. Annu Rev Phytopathol. 42(51):61–82. https://doi.org/10.1146/annurev.phyto.42.040803.140427.
Johnny’s Selected Seed. 2021. Defiant PhR. https://www.johnnyseeds.com/vegetables/tomatoes/slicing-tomatoes/defiant-phr-organic-f1-tomato-seed-2525G.html. [accessed 21 May 2021].
Klee HJ, Tieman DM. 2013. Genetic challenges of flavor improvement in tomato. Trends Genet. 29(4):257–262. https://doi.org/10.1016/j.tig.2012.12.003.
Lammerts van Bueren ET, Myers JR. 2012. Organic crop breeding. Wiley-Blackwell, Ames, IA, USA. https://doi.org/10.1002/9781119945932.
Lenth R. 2023. emmeans: Estimated marginal means, aka least-squares means. R package version 1.8.9.
Lyon A, Silva E, Zystro J, Bell M. 2015. Seed and plant breeding for Wisconsin’s organic vegetable sector: Understanding farmers’ needs. Agroecology and Sustainable Food Systems. 39(6):601–624. https://doi.org/10.1080/21683565.2015.1017786.
Merscher P. 2020. Flavor evaluation for crop scientists: Examining new methods for local food markets (MS Thesis). University of Wisconsin–Madison.
Sacco A, Di A, Nadia M, Trotta N, Punzo B, Mari A, Barone A. 2013. Quantitative trait loci pyramiding for fruit quality traits in tomato. Mol Breed. 31(1):217–222. https://doi.org/10.1007/s11032-012-9763-2.
Seed to Kitchen. 2016. 2016 vegetable variety screening trials: Tomato. Department of Plant and Agroecosystem Sciences, University of Wisconsin–Madison, Madison, WI, USA. www.seedtokitchen.horticulture.wisc.edu/trial-results. [accessed 1 Dec 2023].
Seed to Kitchen. 2015. 2015 vegetable variety screening trials: Tomato. Department of Plant and Agroecosystem Sciences, University of Wisconsin–Madison, Madison, WI, USA. www.seedtokitchen.horticulture.wisc.edu/trial-results. [accessed 1 Dec 2023].
Shelton AC, Tracy WF. 2015. Recurrent selection and participatory plant breeding for improvement of two organic open-pollinated sweet corn (Zea mays L.) populations. Sustainability. 7(5):5139–5152. https://doi.org/10.3390/su7055139.
Souza LM, Melo PCT, Luders RR, Melo AMT. 2012. Correlations between yield and fruit quality characteristics of fresh market tomatoes. Hortic Bras. 30(4):627–631. https://doi.org/10.1590/S0102-05362012000400011.
Sudermann MA, McGilp L, Vogel G, Regnier M, Jaramillo AR, Smart CD. 2022. The diversity of Passalora fulva isolates collected from tomato plants in U.S. high tunnels. Phytopathology. 112(6):1350–1360. https://doi.org/10.1094/PHYTO-06-21-0244-R.
Sun YD, Liang Y, Wu JM, Li YZ, Cui X, Qin L. 2012. Dynamic QTL analysis for fruit lycopene content and total soluble solid content in a Solanum lycopersicum × S. pimpinellifolium cross. Genet Mol Res. 11(4):3696–3710.
Tieman D, Bliss P, McIntyre LM, Blandon-Ubeda A, Bies D, Odabasi AZ, Rodríguez GR, Van Der Knaap E, Taylor MG, Goulet C, Mageroy MH, Snyder DJ, Colquhoun T, Moskowitz H, Clark DG, Sims C, Bartoshuk L, Klee HJ. 2012. The chemical interactions underlying tomato flavor preferences. Curr Biol. 22(11):1035–1039. https://doi.org/10.1016/j.cub.2012.04.016.
Tieman D, Zhu G, Resende MFR, Lin T, Nguyen C, Bies D, Rambla JL, Beltran KSO, Taylor M, Zhang B, Ikeda H, Liu Z, Fisher J, Zemach I, Monforte A, Zamir D, Granell A, Kirst M, Huang S, Klee H. 2017. A chemical genetic roadmap to improved tomato flavor. Science. 355(6323):391–394. https://doi.org/10.1126/science.aal1556.
US Department of Agriculture. 2021a. 2021 agriculture census. https://quickstats.nass.usda.gov/. [accessed 17 May 2021].
US Department of Agriculture. 2021b. Nutrients present in roma tomatoes. https://fdc.nal.usda.gov/fdc-app.html#/food-details/1750354/nutrients. [accessed 17 May 2021].
US Department of Agriculture. 2005. Shipping point and market shipping point and market inspection instructions for tomatoes. https://www.ams.usda.gov/sites/default/files/media/Tomato_Inspection_Instructions%5B1%5D.pdf. [accessed 17 May 2021].
Wang S, Qiang Q, Xiang L, Fernie AR, Yang J. 2023. Targeted approaches to improve tomato fruit taste. Hortic Res. 10(1):uhac229. https://doi.org/10.1093/hr/uhac229.
Wilkerson ED, Anthon GE, Barrett DM, Sayajon GFG, Santos AM, Rodriguez-Saona LE. 2013. Rapid assessment of quality parameters in processing tomatoes using hand-held and benchtop infrared spectrometers and multivariate analysis. J Agric Food Chem. 61(9):2088–2095. https://doi.org/10.1021/jf304968f.
Zhao J, Sauvage C, Zhao J, Bitton F, Bauchet G, Liu D, Huang S, Tieman DM, Klee HJ, Causse M. 2019. Meta-analysis of genome-wide association studies provides insights into genetic control of tomato flavor. Nat Commun. 10(1):1534. https://doi.org/10.1038/s41467-019-09462-w.
