Vegetable grower survey.

An invitation was mailed to U.S. fresh-market vegetable growers using physical addresses obtained from MarketMaker (http://foodmarketmaker.com/) in Jan. 2012. The U.S. fresh-market vegetable growers were invited to participate in an online survey to identify barriers constraining the adoption of organic farming practices by them (Veldstra, 2012). The survey was designed with and hosted in the online-based software Qualtrics (Qualtrics LLC, Provo, UT). As part of this survey, tomato growers were asked to select three of their most desired traits for inclusion in a breeding program. They were also asked to identify their most problematic plant diseases, because high rainfall and humidity in the Midwest region often makes diseases one of the most limiting production factors. For the purposes of this study, data from the survey were confined to growers in the Midwestern region, comprising Illinois, Indiana, Iowa, Michigan, and Ohio, and separated into those using organic (certified and noncertified) (n = 154) and conventional (n = 135) production practices.

Germplasm panel.

A diverse set of tomato germplasm expected to include desirable traits and perform well under organic management was selected in consultation with public tomato breeders and local organic growers (Table 1). The panel included popular heirloom cultivars; “early modern” OP cultivars; F1 hybrids; a wild tomato relative (Solanum pimpinellifolium) with resistance to LB caused by (P. infestans) and EB caused by (Alternaria solani); a susceptible control cultivar susceptible to LB, EB, and SLS caused by (Septoria lycopersici); and segregating breeding populations derived from crosses of parent germplasm with desirable traits. We define early modern as cultivars selected during the 1940–60s when many new university-trained plant breeders were actively selecting improved OP cultivars, and before when breeding efforts transitioned to development of F1 hybrids. Some of the breeding populations were made using two North Carolina State University (NCSU) inbred with high levels of disease resistance and two early modern OP cultivars with excellent flavor and moderate levels of disease resistance. The NCSU inbred lines included NC 1 CELBR and NC 2 CELBR, which have resistance to EB, LB (Ph-2, Ph-3), verticillium wilt (Ve), and fusarium wilt (I, I-2) in a determinate plant (Gardner and Panthee, 2010). The two early modern OP cultivars, Wisconsin 55 and Crimson Sprinter, both have excellent flavor, but lack problems often associated with heirloom cultivars such as fruit cracking, checking, or cat-facing. ‘Wisconsin 55’ has a moderate level of EB resistance (Walker et al., 1948) and has exhibited a degree of polygenic resistance to LB (Gevens et al., 2010). ‘Crimson Sprinter’ has moderate level of resistance to SLS and increased lycopene content due to the crimson gene (og^c) (Graham, 1970). Septoria leaf spot resistance is rare in S. lycopersicum germplasm, and resistance in ‘Crimson Sprinter’ was likely enhanced by over 20 years of selection under disease pressure in North Dakota. Both OP parents have strong indeterminate growth, which tend to have a high foliage to fruit ratio, and generally imparts a higher amount of fixed sugars in the fruit resulting in improved flavor (R. Gardner, personal communication). The other set of breeding populations were made using ‘Richter’s wild currant’ and ‘Matt’s wild cherry’. Richter’s wild currant (S. pimpinellifolium) is resistant to LB and is the original source of the Ph-2 gene in the NC 1 CELBR and NC 2 CELBR inbred lines (Brusca, 2003). It has also shown a moderate level of resistance to EB and SLS. ‘Matt’s wild cherry’ (S. lycopersicum var. cerasiforme) is a Mexican landrace collected from a location near the Toluca Valley (a major center of diversity for P. infestans), and has a moderate level of resistance to LB, EB, and SLS, and has excellent flavor. The genetic basis of the LB resistance is currently unknown, and preliminary marker screens have not detected Ph-2 or Ph-3, although further investigation is underway (E. Haga, Johnny’s Selected Seeds, personal communication). Seed was obtained from tomato breeders, local growers, and commercial seed companies. Certified organic seed was used when possible, and none of the seeds were treated.

Table 1.

Tomato entries planted in germplasm evaluation trials grown under organic management in 2011 and 2012.

View Table

Field trials.

Seeds were planted in 72-cell flats filled with Sunshine^® Natural & Organic Planting Mix. Flats were held under intermittent mist to facilitate germination and moved to the greenhouse after about 2 weeks where they were fertilized with a soluble organic product (Organic Hydrolyzed Fish Fertilizer; Neptune’s Harvest, Gloucester, MA). Eight weeks after seeding, trays were moved outside to a protected area to acclimate for 1 week before transplanting in field trials.

Field trials were conducted at Purdue’s Meigs Horticulture Research Farm south of Lafayette, IN on land managed using certified organic production practices for the past 10 years. Average high and low temperatures at this location range between 26–29 °C and 13–18 °C during July through September. Rainfall averages 101, 88, and 63 mm in July, August, and September, respectively. Trials were conducted on Drummer soil (fine, silty, mesic Typic Haploquoll) that had previously been planted to winter wheat (Triticum aestivum) in 2010, and a sorghum × sudangrass (Sorghum ×drummondii) cover crop followed by a fall planted Austrian winter pea (Pisum sativum) cover crop in 2011. In spring 2012, the cover crop was mowed and disked in about 1 month before tomato transplanting. Cover crop biomass was collected before termination to estimate biomass and potential nitrogen contribution. Soil cores were collected to depth of 30-cm, pooled and submitted to Midwest Laboratories (Omaha, NE) for a basic soil nutrient test (Table 2). Organic fertilizers were applied before planting based on spring soil tests and estimated nitrogen contribution of the winter cover crop in 2012. In 2011, 3–3–3 composted chicken manure (Rose Acre Farms, Seymore, IN) was broadcast at a rate of 6726 kg⋅ha⁻¹, and 3–3–3 Re-Vita Compost Plus (Earth Food, Hartville, OH) was broadcast at a rate of 4484 kg⋅ha⁻¹ in 2012. Fertilizers were incorporated to a depth of 5 to 8 cm before establishment of raised beds that were overlain with drip tape and black plastic mulch. Beds were spaced 2.44 m apart and tomato plants were spaced 0.61 m apart within beds. Tomatoes were transplanted on 19 May 2011 and 31 May 2012 using a water-wheel transplanter. The experimental design was a randomized complete block with four replicates in 2011 and three replicates in 2012. An experimental unit was a 3.66-m long bed with six plants. Tomato plants were pruned to the first flower cluster and periodically tied using the Florida weave design. The plots did not receive any supplemental fertilizer or pesticides in either year. Organic growers were participated in assessment of the performance of the tomato entries during field days and informal farm visits.

Table 2.

Basic soil tests before planting organic field trials in 2011 and 2012.

View Table

Vigor, maturity, and yield.

Plots were assigned an overall vigor rating using a 1–9 scale at least three times during each growing season. The scale was based on plant height, width, and foliar density, where a “1” would be assigned to a small plant with a sparse canopy, and a “9” would be assigned to a large plant with dense canopy. Fruit was not harvested from the very small fruited entries because of the substantial amount of labor that would have been required. Fruit was harvested from each plot in all of the other entries at weekly intervals once plants started to bear ripe fruit. Fruit that was ripe or just past breakers was collected during each harvest event, separated into marketable and nonmarketable categories, and counted and weighed. Nonmarketable fruit had severe blemishes and/or pathogen lesions that would have prevented sale in local markets; all other fruit was considered marketable. After a hail storm in early Aug. 2012, damaged fruit was removed from all plots, and the weight of damaged fruit was not included in the yield estimates.

Disease.

The identity of foliar pathogens infecting tomato plants was confirmed by the Purdue Diagnostic Laboratory in each year. Total leaf defoliation by all foliar pathogens was used to quantify disease severity in each plot. Disease ratings were initiated once symptoms began to appear, and plots were rated monthly until the first frost. The Horsfall–Barratt scale was used to quantify disease severity with the midpoints used to calculate the area under the disease progress curve (AUDPC) using the formula presented by Shaner and Finney (1977).

Hornworms.

Hornworm (M. sexta) defoliation was quantified in September of each year after significant hornworm pressure had occurred. Defoliation was determined visually using a 1–100 scale for percentage of the total plot area defoliated.

Statistics.

All data were analyzed using SAS software (Version 9.3; SAS Institute Inc., Cary, NC). Data from the survey were analyzed using the chi-squared (χ²) test. The χ² test was performed using the FREQ procedure to create a 2 × 13 contingency table to test the null hypothesis of no association between producer group and 13 traits included in the survey. Statistical differences between organic and conventional tomato producers were detected by constructing 95% confidence intervals for the difference between two proportions. Data from the field trials were analyzed using analysis of variance (ANOVA) (PROC GLM), and mean separation conducted using Fisher’s least significant differences test. The Bonferroni correction was applied to account for multiple comparisons. ANOVA included tomato entry, entry type, year, and their interactions as sources of variance. Data were checked for model assumptions, and square root or log transformed when normality or equality of variance were not met. Data were back-transformed to report means. Relationships between AUDPC, hornworm defoliation, and each of the harvest characteristics were determined using spearman rank correlation analyses using PROC CORR. Data are reported separately by year because of significant year by entry interactions, with the exception of the hornworm data where the interaction of these factors was not significant.

Phase 1—identifying growers’ needs.

The first phase of a participatory breeding program is to identify growers’ needs. One of the conundrums of plant breeding is that it can often drive production systems and markets, i.e., growers and consumers might not know they want something until they see it, making it difficult to accurately capture needs and identify targets for breeding programs. We used an online survey to identify key traits that Midwest tomato growers believe are key to optimizing organic tomato production. Results were separated into those using organic and conventional production systems to determine whether separate breeding programs are warranted. Systematic studies conducted via grower surveys can provide valuable information about the relative importance of key traits, and increase the applicability of varietal development efforts (Yue et al., 2013).

Organic (79.2%) and conventional growers (72.6%) ranked flavor as their top breeding priority and did not differ from each other in the importance of this trait (Table 3). The grower groups differed, however, in the importance placed on other fruit quality characteristics. Conventional growers were more concerned with characteristics related to appearance, with 56.3% rating crack resistance and 28.1% rating color/shape as important, whereas only 38.3% and 18.2% of organic growers rated these traits as important, respectively. In contrast, organic growers were more concerned with nutritional quality than conventional growers, with 27.3% rating this as a priority in comparison with just 4.4% of conventional growers. These data are not surprising given that concerns about cosmetic surface blemishes are less important to customers who purchase organic produce (Goldman and Clancy 1991), whereas “health” is often noted to be the primary reason consumers buy organic food with many believing that organic food is more nutritious (Hughner et al., 2007). Nevertheless, the relatively high importance expressed by both grower groups indicated that fruit quality traits should be an important component of fresh-market tomato breeding programs.

Table 3.

Percentage of Midwest conventional (CO) and uncertified organic and certified (OR) vegetable growers indicated that a specific trait was one of their three most desired traits for inclusion in a tomato breeding program.

View Table

Among agronomic traits, disease resistance ranked as one of the top breeding priorities among both grower groups. However, conventional growers (83.0%) were more likely to rate disease resistance as a priority than organic growers (62.3%) (Table 3). This was surprising given that organic growers have fewer options to control diseases once outbreaks occur. Conventional growers were also more likely to rate SLS, bacterial spot (Xanthomonas campestris pv. vesicatoria), bacterial speck (Pseudomonas syringae pv. tomato), bacterial canker (Clavibacter michiganensis pv. michiganensis), and gray mold (Botrytis cinerea) as a pest than organic growers, while organic growers were more likely to consider white mold (Sclerotinia sclerotiorum) and root-knot nematodes (Meloidogyne spp.) as difficult to control (Table 4). These findings will help tomato breeders to target pathogens that are most relevant to conventional and organic production systems, but they also raise interesting questions about why disease and individual pathogens might be perceived to be more problematic in one system relative to another. Previous studies have indicated that some diseases are less severe in organic vs. conventional tomato farms (Workneh and van Bruggen, 1993), and soil collected from organic farms has been found to be more suppressive to pathogen development than soil collected from conventional farms (Liu et al., 2007). Organic growers commonly apply compost and other biologically based amendments, which can contribute to the development of disease suppressive soil (Bonanomi et al., 2010). Alternatively, organic growers may be more tolerant of certain diseases as indicated by the lower importance placed on appearance characteristics. Additional research to identify the most problematic diseases and mechanisms contributing to disease severity in these systems will help both groups better manage plant diseases.

Table 4.

Percentage of Midwest uncertified organic and certified organic (OR) and conventional (CO) growers who identified each disease as a pest in tomatoes and indicated that the disease was difficult to control based on their responses to an online survey.

View Table

Organic and conventional growers both ranked the most problematic diseases in a similar manner (Table 4), with the foliar diseases LB, EB, and SLS identified as the top three pests. These pathogens have previously been noted to be particularly problematic in tomato. Late blight spreads rapidly under ideal environmental conditions and can completely decimate an entire crop within 7 to 10 d (Nowicki et al., 2012). Early blight can completely defoliate tomato plants resulting in yield reductions of more than 79% (Foolad et al., 2008). Septoria leaf spot can cause severe defoliation and crop loss, particularly in humid regions during periods of high rainfall and frequent dew (Delahaut and Stevenson, 2004). Conventional growers can turn to synthetic pesticides to help manage foliar pathogens, though rapid spread under ideal environmental conditions and potential for development of fungicide resistant isolates are making these diseases increasingly difficult to manage. Organic growers commonly rely on crop rotation to disrupt pathogen lifecycles and reduce disease outbreaks. However, while this approach has been successfully used to suppress many soil-borne pathogens, it is far less effective for foliar pathogens that can blow in from neighboring farms and/or natural areas. Organic fungicides containing copper provide fair control of foliar pathogens, but these products are contact, not systemic, and must be applied often resulting in buildup of residues that negatively impact soil and water quality (Eijsackers et al., 2005). Copper fungicides are now excluded from organic production systems in Europe, and are expected to be banned under U.S. national organic standards soon. Consequently, planting disease-resistant cultivars is the most valuable tool for managing foliar pathogens like these in both production systems.

Both grower groups ranked weed competitiveness relatively high, but did not differ in the importance of this trait (Table 3). This was surprising given that organic growers often cite weed control as one of the biggest factors preventing them from transitioning to organic production (Veldstra, 2012; Walz, 2004), and weed competiveness has been identified as a top breeding priority in other organic crops (Lammerts van Bueren et al., 2011). Genetic variation to weed competiveness has previously been observed in tomato (Ngouajio et al., 2001), indicating that this is a trait that could be selected for in future breeding programs.

The relatively low importance placed on breeding for weed competitiveness in our survey could be due, at least in part, to the fact that Midwest tomato growers commonly rely on black plastic mulch to help control weeds. However, black plastic mulch cannot be reused and generally ends up in landfills, which has stimulated interest in alternatives like the use of no-till cover crop systems to help control weeds in organic vegetable systems (Leavitt et al., 2011; Mulvaney et al., 2011).

Phase II—selecting suitable material to test with growers.

The second phase of a participatory breeding program is to select suitable material to test with growers. The material must be evaluated to ensure that desirable traits are expressed under low-input conditions that characterize organic systems, because this cannot be guaranteed if selection was conducted under high-input production systems (Lammerts van Bueren et al., 2011). Preliminary evaluations like these also allow breeders to better understand agronomic constraints, and eliminate germplasm with limited applicability for organic systems. Our preliminary germplasm trial included a diverse set of existing cultivars and F1 hybrids expected to perform well in organic production systems, as well as experimental breeding populations developed specifically for organic varietal development efforts.

Foliar diseases were problematic in both years, and EB was the most common disease observed. Disease incidence across entries was less severe in 2012 than 2011, and corresponded with severe drought conditions experienced in 2012 (Table 5). Disease incidence varied widely among entries regardless of entry type, and there were significant year × entry interactions (Table 5). We did not differentiate between foliar diseases in our ratings, which may account for some of the variability between entries across the 2 years. However, some entries exhibited relatively consistent disease pressure in both years. For example, the susceptible control ‘Silvery Fir Tree’ and the heirloom ‘Arkansas Traveler’ exhibited relatively high disease incidence in both years, whereas the F1 hybrid ‘Mountain Magic’ and several experimental breeding populations including ‘OSA811’, ‘OSA823’, and ‘OSA2020-2’ exhibited relatively low disease incidence in both years. ‘Mountain Magic’, ‘OSA811’, and ‘OSA823’ were all derived from NCSU inbreds that hold quantitative resistance to EB as well as LB resistance derived from the single genes Ph-2 and Ph-3. ‘OSA2020-2’ was also derived from material expected to confer resistance to EB and LB. Late blight was not observed in either year of the preliminary germplasm trial, however, there was a severe outbreak of LB at our research farm in 2013, and these breeding populations exhibited significant resistance in comparison with a susceptible control (Hoagland, unpublished data). These results confirm that selecting for resistance to foliar pathogens, particularly EB, will be key to optimizing productivity in Midwest organic production systems. Germplasm with durable resistance to these pathogens appears to be present in several of the experimental breeding populations, and further selection is warranted.

Table 5.

Area under the disease progress curve (AUDPC) for leaf defoliation by foliar pathogens in 21 tomato entries grown under organic conditions at Purdue’s Meigs Research Farm south of Lafayette, IN.

View Table

Insect pests were not particularly severe in either year, however, hornworms (M. sexta) were present in both years and caused significant damage in some entries (Fig. 1). Interestingly, there were wide differences among entries in susceptibility to hornworm defoliation regardless of entry type, and there was not an entry × year interaction. Several experimental breeding populations including ‘OSA823’, ‘OSA801’, ‘OSA2020-2’, the F1 hybrid ‘Frazier’s Gem’, and our susceptible control ‘Silvery Fir Tree’ (an early-modern OP) exhibited relatively low defoliation. A wild tomato relative has previously been documented to exhibit potential resistance to hornworm and other insects in response to chemical compounds released via exudates from trichomes on the leaf surface (Kennedy and Yamamoto, 1979). However, breeding for insect resistance has largely been neglected in most tomato breeding programs because of difficulty in phenotypic screening for resistance in field trials, problems associated with linkage drag from wild relatives, and the ease of control by pesticides (Foolad, 2007). Identifying germplasm with hornworm resistance and uncovering the mechanisms that mediate this resistance will help breeders begin to select for this trait. Some of the experimental breeding populations exhibited potential resistance to both hornworms and foliar pathogens, but there was no correlation between these two traits indicating that different mechanisms operated to suppress these pests (data not shown).

View Full Size

Percent hornworm defoliation in 22 tomato entries grown under organic conditions at Purdue’s Meigs Farm south of Lafayette, IN, and averaged over 2011 and 2012 growing seasons.

Citation: HortScience 50, 9; 10.21273/HORTSCI.50.9.1301

• Download Figure
• Download figure as PowerPoint slide

There were wide variations among entries in harvest characteristics in both years regardless of entry type, and significant year × entry interactions (Tables 6 and 7). However, there were no strong correlations between disease or pest incidence and any of the harvest characteristics (data not shown). Marketable fruit yield across entries was lower in 2012 than 2011, and corresponded with severe drought conditions and a hail storm that damaged fruit in 2012 (Tables 6 and 7). The small- to mid-sized fruited F1 hybrid ‘Mountain Magic’ exhibited the greatest total marketable fruit yield, and had high percent marketable fruit in both years. Other small fruited entries, including the heirloom ‘Tainin’, and the early modern ‘Green Doctor’, also had relatively high total marketable fruit yield and a low percentage of cull fruit in both years. Small fruit is generally less susceptible to cracking and other physiological disorders like cat-facing that negatively impact fruit quality. However, in a consumer preference study, Oltman et al. (2014) found size to be one of the most important attributes in tomato, with the ideal size being medium baseball-sized fruit. Growers who participated in assessment of the entries in our trial during farm visits and field days confirmed that they prefer medium-sized fruit because it is generally more profitable.

Table 6.

Average marketable and cull fruit weight and number of 18 tomato entries grown under organic conditions in 2011 at Purdue’s Meigs Horticulture Research Farm south of Lafayette, IN.

View Table

Table 7.

Average marketable and cull fruit weight and number of 18 tomato entries grown under organic conditions in 2012 at Purdue’s Meigs Horticulture Research Farm south of Lafayette, IN.

View Table

Much of the resistance to EB and LB in tomato is derived from the small fruited wild relatives of S. pimpinellifolium and S. lycopersicum var. cerasiforme, which make selecting for both disease resistance and larger fruit size challenging. These wild relatives also tend to be highly vegetative (high foliage to fruit ratio) resulting in low fruit yield (E. Haga, Johnny’s Selected Seeds, personal communication). The experimental breeding populations had medium-sized fruit and average yield in 2011, but their yield was relatively low in 2012. Yield is generally considered an important factor in most production systems, and increasing yield remains a universal standard in tomato breeding programs (Foolad, 2007). Ongoing selection efforts will need to focus on selecting individual plants with high yield from among the segregating breeding populations. In addition, the breeding populations may need to be backcrossed with cultivars like Arkansas Traveler, which had relatively high marketable yield and percent marketable fruit in both years of our trial.

We did not quantitatively measure fruit quality, including flavor and nutritional content in the preliminary germplasm trial because of the time and expense associated with these analyses. However, the experimental breeding populations were developed with these characteristics in mind, and selection for these traits will need to be an explicit component of future selection efforts in on-farm trials. Genetic variation for tomato’s nutritional quality has been established, particularly for carotenoid (Levin et al., 2006) and flavonoid (Mes et al., 2008) compounds, making selection for this trait fairly straight forward. In contrast, selection for flavor is likely to be more difficult. Qualities that influence the perception of tomato flavor are well known, including taste, smell, texture, appearance, and mouth feel. However, flavor is one of the most difficult traits to breed for, because it is highly subjective and growers and consumers often disagree on what good flavor implies (Lammerts van Bueren et al., 2011). Tomato quality traits are also highly influenced by genotype × environment interactions, with some cultivars exhibiting as much as a 211% difference in fruit quality characteristics in response to planting location (Panthee et al., 2012), highlighting the need to select in target crop environments. While such differences create a challenge for tomato breeders, they also represent an opportunity for developing niche-specific cultivars with improved fruit quality (Panthee et al., 2012). Alternatively, breeders could select for stable flavor across locations to mitigate potential genotype × environment interactions. Regardless of the approach, analytical chemistry and developing high throughput protocols using infrared spectroscopy could help breeders select for flavor by identifying balanced levels of acids and sugars, as well as specific volatile compounds previously identified as key elements of flavor (Buttery, 1993). Sensory analysis performed by trained taste panelists is another method that could be used to select tomatoes with superior flavor (Rocha et al., 2013).