Improving farmer options for sustainable and profitable direct-market tomato production and high tunnel management in the Upper Midwest

1. Comparing high tunnel, caterpillar tunnel and field production

Complete results are available in the Horticulture Masters thesis of Juan Astroza, available from the University of Wisconsin Library System or by request.

Environmental Conditions

Figure 1 shows the maximum daily temperature for each system in Spooner. Madison results were very similar. In both Madison and Spooner, caterpillar tunnels and high tunnels had higher daily maximum temperatures compared with the open field, at times reaching a difference of 15 ˚C or more, particularly at the end of the season. Conversely, minimum daily temperatures were very similar for all systems and in both locations.

Table 2 shows the average maximum and minimum daily temperature, growing degree day accumulation and date of first harvest for the three management systems in each location. High tunnels and caterpillar tunnels had about 5˚C higher average maximum daily temperature compared to the open field, but there were not strikingly different average minimum temperatures between the tunnels and the open field.

To calculate the growing degree day accumulation, the maximum temperature for each day was capped at 30˚C as exceeding this temperature has detrimental effects on plant growth (Pathak, 2018). The base temperature used was 10˚C and if there was a negative result of daily GDD accumulation it was simply marked as zero. GDD started to accumulate when the plants were transplanted into the tunnels or field. As shown in Table 2, the first harvest from the high tunnel and caterpillar tunnel for Madison was July 17^th. At that moment, the plants had accumulated 1204.1 and 1186 GDD respectively. The first harvest from the open field at Madison was August 20^th with a GDD accumulation of 1283.9. The accumulation of GDD in Spooner at the moment of the first harvest was 1237 for the high tunnel, 1185 for the caterpillar tunnel and 1285 for the open field.

Production

Table 3 shows the means by trait for each management system and location for each of the traits of interest in the experiment. Significance groupings for pairwise comparisons between systems and locations are given. Tests were done separately for management system by location values for each trait, and for main effects of management and location for each trait.

Cells with the same font (regular, italicized or bold) were compared to each other within a trait and management systems that do not share a letter are significantly different at the p<0.05 level.

Marketable yields:

Management had a significant effect on marketable yields (p-value < 0.001). All management systems were significantly different from each other as seen in Table 3. Location was also significant (p-value 0.0014), with significant differences between Madison and Spooner recorded. There was a significant interaction between management and location (p-value <0.001). When comparing management systems between locations, the open field was significantly worse than the tunnels at both locations, but the reduction was greater at Madison and the relative performance of the tunnels also changed. The caterpillar tunnel was not significantly different from the high tunnel at Madison but did have significantly better marketable yield than the high tunnel at Spooner (Figure 2). There was evidence for a location and year interaction (p-value <0.001) but the ranking of the three management systems was relatively consistent within location across years. The caterpillar tunnel outperformed the high tunnel in Spooner in 2018, which drives the difference between the caterpillar tunnel and high tunnel in that location. There was also strong evidence for a management and year interaction (p-value <0.001). This was also influenced by the performance of the caterpillar tunnel in Spooner in 2018, which was significantly higher yielding than the high tunnel compared to the two subsequent years where the performance was more similar. The field was always the worst in terms of marketable yield. Across locations, there was not a clear trend between performance in the caterpillar tunnel and high tunnel, and the two systems were not usually significantly different from each other.

Variety had a significant effect on marketable yield (p-value < 0.001). The interactions between variety and management and variety by location were both significant (p-values < 0.001). However, this interaction was primarily scalar in nature, with very few changes in variety rankings. Figure 3 shows marketable yield by variety, system and location.  There were no significant changes in ranking of varieties across the three systems within a location, and there was only one significant change in rank across locations, with Defiant as having the highest marketable yield in Madison and Big Beef having the highest marketable yield in Spooner. There was a significant interaction between variety, management and location (p-value: 0.0018), but consistency in variety rankings. Damsel and Defiant seem to have a larger increase in performance in the caterpillar tunnel than in the high tunnel or the field compared to other varieties in Madison, however the magnitude of crossover interactions is very small. Similarly, there was a significant interaction of variety by location by year by management (p-value 0.021), but there were very few crossover interactions within a location or within a management system, with the exception of JBT having significantly higher marketable weight in the field in Spooner in 2019.

All varieties had higher performance in the tunnels and variety recommendations can be made for each location without having to make recommendations specific to management systems. Generally, marketable count followed the same patterns as marketable weight. Defiant had significantly higher marketable fruit count than the rest of the varieties and Paul Robeson and Pruden’s Purple had the lowest fruit count.  For average fruit weight, the open field had smaller fruit than both tunnels, with no significant difference between the caterpillar tunnel and high tunnel. We also saw significant evidence for the location main effect (p-value <0.001), with Spooner having significantly higher average fruit weight than Madison. The interaction between management and location was significant (p-value 0.032) but there were no significant differences between management systems within a location.

Management system had a significant effect on the proportion unmarketable by weight (p-value <0.001). The open field had the highest proportion unmarketable, followed by the caterpillar tunnel and the high tunnel. There were also differences in causes of unmarketability. In the top causes recorded per harvest, blossom end rot as a specific problem was more prevalent in the high tunnel, while hornworm damage was more prevalent in both caterpillar tunnel and high tunnel than in the open field. For causes common to all systems including stem side crack, radial splitting, the frequency of the problem being listed in the top three was the same, but the field had more unmarketable as a proportion of total harvested. The main effect of location was significant and showed that Madison had a significantly higher proportion of unmarketable yield than Spooner. This was primarily due to higher levels of stem side cracking, blossom end rot and hornworm damage.

Variety also had a significant effect on the proportion unmarketable (p-value <0.001). Defiant had the lowest proportion of unmarketable weight across management, location and year. In terms of the causes of unmarketability, blossom end rot was highest in Damsel and Japanese Black Trifele and it was lowest in Paul Robeson and Defiant. Stem side cracking showed no difference between varieties except for Defiant, which was significantly lower than the rest. For radial splitting Defiant was also significantly lower than the rest, and Paul Robeson and Pruden’s Purple were significantly higher than the other varieties. The effects of the interactions between genotype and management were significant (p-value <0.001). All varieties had a consistently lower proportion of unmarketable weight in the high tunnel, then caterpillar tunnel, and the highest proportion of unmarketable weight in the open field, with the exception that Pruden’s Purple and Defiant both had no significant difference between management systems in Spooner. Rankings were consistent over years.

Disease incidence and severity:

The open field had significantly higher total area under the disease progress curve (TAUDPC, all diseases combined) followed by the high tunnel, then the caterpillar tunnel. Figure 4 shows the TAUDPC in Madison across the three systems.  The high tunnel may have had higher levels of humidity than the caterpillar tunnels, leading to higher levels of leaf mold in the high tunnel.  Field disease pressure was primarily septoria leaf spot and early blight, and in the two tunnels disease pressure was primarily early blight and leaf mold.  The reduction in septoria leaf spot in the two tunnel systems was particularly striking. Table 4 gives means by management systems for disease and quality traits. There did not seem to be a strong separation of varieties based on market classes. Big Beef in the high tunnel had a significantly higher TAUDPC than all other varieties except Pruden's Purple (Figure 5).

Septoria Leaf Spot is one of the most critical diseases for organic tomato production. In Madison, the main effects of year (p-value <0.05) and management (p-value <0.05) were significant. The interaction of year by management is also significant (p-value <0.05). When Spooner is added to the analysis and compared for 2020 only the management main effect is significant (p-value <0.05). There were no significant differences among varieties. The caterpillar tunnel and high tunnel have almost no disease, but the open field is significantly higher than both tunnel systems.

Fruit Quality:

Management had a significant effect on ˚Brix levels (p-value < 0.05). The high tunnel was not significantly different than the open field and caterpillar tunnel, but the caterpillar tunnel had significantly higher ˚Brix levels than the open field in 2020 for both locations. In Madison across years, both tunnels had significantly higher ˚Brix than the open field but there was no difference between the tunnels (figure 7). The location had a significant effect on ˚Brix levels (p-value < 0.05) with Madison having significantly higher ˚Brix levels compared to Spooner in 2020. In Madison, there was strong evidence of the interaction effect of year by management on ˚Brix levels (p-value < 0.05). However, there were not crossover interactions among management systems at WMARS across years. There was no significant interaction between location and management in 2020.

Variety also had an impact on ˚Brix levels (p-value <0.05) but pairwise comparisons did not show many significant differences between varieties (Figure 8). Damsel was consistently a top performer across all management systems, years and locations. Paul Robeson was also in the top performers for the caterpillar tunnel and open field but not for the high tunnel. Damsel performed exceptionally well in high tunnels for both locations in 2020. There was no clear worst performer. Big Beef, Defiant and Pruden’s Purple had a steep decline in the open field. Paul Robeson and Damsel also saw a decline in the open field compared to their performance in the caterpillar tunnel, and to the high tunnel for Damsel, but maintained reasonably high ˚Brix levels compared to other varieties.

Management also had an effect on citric acid (CA) levels (p-value <0.05). Across locations and for year 2020, there was not a significant difference in CA levels between the caterpillar tunnel and open field, but the high tunnel had significantly lower levels than the other two systems. In Madison across three years, there was no difference between the tunnels, but they were both significantly lower than open field(Figure 9). Location had a significant effect on CA levels (p-value <0.05). For year 2020, WMARS had significantly higher levels of CA compared to Spooner. There was some evidence of the interaction effect of location by management on CA levels (p-value: 0.072), but no crossover interactions among management systems. Variety had a significant effect on CA levels (p-value <0.05). For year 2020, across both locations, there was a significant difference between varieties on CA levels. The interaction of variety by management was significant for year 2020 only (p-value < 0.05). There was a trend for almost all varieties except Defiant and Japanese Black Trifele of having higher CA levels in open field compared to both tunnels.

For the tasting data, management had a significant impact on on bitterness, acidity, sweetness, texture and overall flavor. When comparing the influence of management systems in acidity, the open field was significantly higher than both tunnels, and there was no significant difference between tunnels. Comparing sweetness, the tunnels were higher than the open field but not significantly different from each other. For overall flavor ratings, just like sweetness, fruit from both tunnels were rated with superior flavor compared to the open field and showed no difference between the tunnels (Figure 9).

Variety had a significant impact on intensity, umami, acidity, sweetness, texture, appearance and overall flavor. Damsel showed the highest scores in overall flavor compared to other varieties and in specific scores like texture, intensity and sweetness was superior to all or most of the other varieties (figure 10). There was not a clear variety that performed worse than the rest. There was a consistent trend of all varieties showing lower overall flavor ratings in the open field compared to both tunnels.

On-farm trials

For both market classes, slicers and heirloom, there was a trend for better performance in terms of productivity and flavor in high tunnel and caterpillar tunnel than in the open field. There was not a clear difference between caterpillar tunnel and high tunnel in terms of performance. In some varieties and for some farmers the caterpillar tunnel had better performance and for other farmers and varieties the high tunnel was better.

Costs of production and revenue potential

The breakdown by task showed that the primary activity by hours of labor requirement in each management system is harvesting with 130 hours for the high tunnel, 132.6 hours for caterpillar tunnels and 60 hours for the open field. The second most important task by hour requirement was trellising with 60 hours for the tunnels and 36 hours for the field. One of the tasks that showed big differences between systems was ventilation, with 7.2 hours for the high tunnel, 54 hours for the caterpillar tunnel with one row, an estimated 27 hours for the caterpillar tunnels with two rows and in the open field there was no ventilation needed. Tasks such as scouting, transplanting, and field cleanup did not show much difference between management systems, whereas pruning and trellising were slightly higher in the tunnels. For the total labor hours required by management system, the highest was the one row caterpillar tunnel with 464 hours; second was the high tunnel with 376 hours; third was the estimated amount for a two-rows caterpillar tunnel with 352.7 hours; and the lowest requirement was the open field with 236 hours.

A cost comparison of the three management systems, including both of the variations of the caterpillar tunnels (one and two rows) was performed (Table 5). These costs include management costs, purchase of the structure, maintenance expenses, and building costs. This was done for the equivalent of one high tunnel fitting 294 plants, and the number of caterpillar tunnels and field space for this number of plants was calculated. The costs of management were calculated with the data from the Madison trials that were recorded in the 2020 season, and US$10 per hour of labor cost was considered for the analysis. The management with the highest cost by season was the one row caterpillar tunnel, followed by the high tunnel and the two row caterpillar tunnel. The management with lowest cost was the open field. The purchase cost was calculated from different sources. For the high tunnels, prices were used from tunnels acquired from Zimmerman’s Welding LLC with a purchase cost of US$13,000 per tunnel including all materials. The caterpillar tunnels purchase costs were sourced from different companies depending on the material. For example, the greenhouse plastic was sourced from Nolts Produce Supplies, the PVC pipe from First Supply, the ground posts from Wiedenbeck, and the steel corkscrew anchor and rope from Farmtek, with a total cost of US$1,500 per tunnel. The open field costs are much lower due to the relatively few materials needed to implement this management system. The primary costs are the t-posts needed for the Florida weave trellis system that is commonly used in the NCR. T-posts can be found in most hardware stores and the cost is about US$3.5 each. The open field of a size that will fit 294 plants will need 112 t-posts, which translates to a total expense of purchase in infrastructure of US$392. Common materials, such as plastic mulch, drip tape, mainline, emitters, and trellising string, among others, for the three systems were not included in the calculations.

For the building costs, person-hours needed for construction were gathered from the Madison estimating a 400 person-hours requirement for the high tunnel, eight person-hours for each caterpillar tunnel and zero building hours for the open field because they are included in the management segment for the moving task each year. Considering US$10 of labor cost, the high tunnel has a building cost of US$4,000. Each caterpillar tunnel of US$80, translating into US$480 for the six tunnels needed for one row caterpillar tunnels and US$240 for the two rows caterpillar tunnels. Adding up management, purchase and building costs, the highest management system cost is predictably the high tunnel, followed by the one row caterpillar tunnel and then the two-row caterpillar tunnel. The cheapest management system is the open field.

A 10-year cumulative net income projection was developed to help farmers understand the differences between management systems over time (Figures 26-29). The projection was calculated by adding up the difference between sales and total expenses for each year. The sales were calculated from the total marketable yield data from each management system for Madison in the years 2018, 2019 and 2020. The average sale price of US$7.7 per kilogram was used as per Hodge et al. (2018) and it was assumed that 100% of the marketable yield was sold at this price. The total marketable weight for the season for each system used for the calculations were the following: 1,480 kilograms for the high tunnel, 1,384 kilograms for caterpillar tunnels and 458 kilograms for the open field. The total expenses for each year included in year 1 were the purchase of the tunnels, building expenses and management costs. Year 4 and year 8 included maintenance costs that consist of the purchase and installation of polyethylene plastic film replacements. All other years included only management costs. The net revenue of each year was added to the previous one, thereby creating a projection of cumulative net revenue. The only management system that started with a negative net revenue in year 1 was the high tunnel, but all three systems had a positive net revenue every year starting in year 2.

The management system with the best performance over the ten years was the two rows caterpillar tunnel recovering the investment on year 1, with year 1 at a cumulative net revenue of US$2,418 and $64,852 cumulative net revenue after ten years, averaging $6,485 yearly net revenue. Caterpillar tunnels with one row had negative net revenue in year one of $3,439 and started accruing a positive cumulative net revenue at year 2 with $2,602. The cumulative net revenue over the ten-year period was calculated at $46,954 and averaged a yearly net revenue of $4,695. The high tunnel had two years of negative cumulative net revenue with -$9,332 and -$1,663 but started having positive cumulative net income at year 3 with $6,005. The total cumulative net income over the ten-year period was $56,281 and the average yearly net income was $5,628. The open field never had a negative yearly net income, but its average yearly net income is $1,140 and totaling a cumulative net income over the ten-year period of $6,092, significantly lower than other management systems.

Discussion

Environmental conditions

One considerable benefit of high tunnel production is the accumulation of more growing degree days that can result in earlier maturity, which confirms previous studies (Both et al. 2007 Waterer and Bantle 2000). Temperature has a direct impact on tomato development and calculating GDD is a useful tool to predict or explain different situations that might help farmers make better decisions. GDD requirements for tomatoes are well studied, although requirements vary slightly between different varieties. High tunnels do this by increasing higher temperatures during the morning until the evening, extending the time plants are in the optimum growth temperature range. Prior research had not yet compared the temperature and degree day benefits of caterpillar tunnels to high tunnels.

Important differences in air temperature were recorded between the three management systems in both locations for the year 2020. Daily maximum air temperature was consistently higher in both tunnels compared to the open field, especially when tunnels were closed, which is mostly early and late in the season. Temperatures remained higher, but with more limited variation in both tunnels in the middle of the summer when tunnels were open. This can be beneficial for tomatoes because a higher temperature will stimulate plant growth and fruit maturity (Both et al. 2007, Waterer and Bantle 2000) but can at the same time be disadvantageous when temperatures rise above 30˚C. Depending on the variety, this high temperature could start inducing flower abortion and this worsens as temperatures increase (Mefferd 2017). In both locations, Spooner and WMARS, and in both tunnels, daily maximum air temperature was recorded above 35˚C several times, compared to the open field that never went above 35˚C, except for the weekend of 4^th of July, which was an exceptionally warm day. It reached 36.2˚C in the open field, 40.5˚C in the caterpillar tunnel, and 40.7˚C in the high tunnel. Research has shown that respiration outpaces photosynthesis at temperatures above 35˚C, which means that the plant is using more energy than it is capable of producing. This process could begin with a decrease in growth. If conditions persist or temperatures go higher, it can lead to serious damage in the plant (Wolfe et al. 1988).

On the other side of the temperature spectrum, minimum daily air temperatures were very similar across management systems in both locations. Surprisingly, temperatures in the open field overnight were slightly higher compared to the tunnels on some nights. This is in agreement with what Zhao and Carey (2009), Rogers and Wszelaki (2012), and Biernbaum (2013). The primary manner that high tunnels with closed sides loose energy during clear nights is by radiation to the cold sky. When tunnels are properly sealed, however, radiant heat from the soil and plants is trapped inside when moisture condenses upon touching the plastic film, just as moisture in atmosphere in cloudy nights reflects the radiant heat from soil back to soil (Biernbaum 2013). However, most high tunnels and caterpillar tunnels are not perfectly sealed loosing moisture and, therefore, are prone to heat loss. Another reason contributing to the slightly lower minimum temperatures in the high tunnels and caterpillar tunnels compared to open fields, and likely the most important one, is that soil temperature in both tunnels was consistently lower compared to open field. This was probably caused by several factors. One factor is that plants in both tunnels were much bigger because they were planted earlier and gave more shade to the soil, keeping the temperature lower. Also, in both tunnel systems, following farmer practices, landscape fabric was used instead of plastic mulch, which may have been another factor that contributed to the lower temperatures given that landscape fabric is permeable and therefore traps less heat during the day.

Similar results can be observed (Table 2) when daily average maximum and minimum air temperature are calculated. Maximum daily average temperatures were about 5˚C higher in caterpillar tunnels and high tunnels compared to the open field in both locations, but average daily minimum temperatures are almost the same between systems and even slightly higher in the open field. This shows that most of the benefits that both caterpillar tunnels and high tunnels bring with respect of temperature is through the elevated temperatures during the day and a slower cooling of temperatures by night, extending the window that the plant is in the optimal growth temperature range, which is usually between 25˚C and 30˚C (Csizinszky 2005). Tomatoes are known for compensating growth when night temperatures are below optimum if temperatures during the day are in the optimum range (Calvert 1964).

Comparing the GDD that plants accumulated in all management systems and locations shows that both tunnels and open field reached physiological maturity at similar GDD to what was expected from Pathak (2018). Although GDD accumulation at first harvest was similar between systems, the dates when that level of GDD was reached differed greatly. At WMARS, the first harvest was July 16^th in both tunnels and August 20^th in the field meaning a 35-day difference between both systems. A smaller difference was observed at Spooner, with the first harvest happening August 2^nd in both tunnels and August 25^th for the open field, with a difference of 23 days. We know that both tunnels were transplanted 34 days before the open field at WMARS and 18 days before at Spooner, and the first harvest in the tunnels was also 35 days before at WMARS and 23 days before at Spooner compared to the open field. Or when looked in days from transplanting to harvest, tunnels at WMARS needed 65 days, versus 66 days for the field. And in Spooner needed 66 days in the tunnels and 71 days in the open field. These results do not show a large difference in days needed between tunnels and open field, but what is important is that the tunnels were planted 34 and 18 days earlier in WMARS and Spooner with lower temperatures from earlier in the season they were still able to produce fruit in 1 day fewer for WMARS and 5 days fewer for Spooner between transplanting and first harvest. In other words, the tunnels were able to accumulate GDD in about the same amount of time or less when outside temperatures were lower. Later in the season, the caterpillar tunnels and high tunnel were mostly closed to increase air temperature. In September, maximum daily temperatures in both tunnels were significantly higher compared to the open field, allowing the plant to be more time in the optimum temperature range of 25-30˚C, keeping the plant producing more, fruit ripening, improving fruit quality and extending the harvest season. The temperature benefit from tunnels will combine with the disease mitigation benefit from tunnels, allowing to keep healthy plants for longer in the tunnels, that will produce more and better fruit quality. There did not seem to be a large difference between the high tunnel and caterpillar tunnel, showing that caterpillar tunnels provide temperature and GDD benefits more similar to high tunnels than the field.

Production traits

Our results show that tomato yields in caterpillar tunnels can be equal to or sometimes even higher than in high tunnels. While there was evidence of an interaction between location and management system, this was consistent over the years in the study, with no significant changes in the rank of management system within location over years. The caterpillar tunnel performed better than the high tunnel in Spooner, largely due to exceptional performance in 2018, and was equivalent to the high tunnel at WMARS for marketable weight. Both systems were higher yielding than the open field, this goes in line with the study performed by O’Connell et al. (2012) that suggests that up to a 33% increase in high tunnel production compared to open field can be achieved if proper management techniques are implemented. Also, Rogers and Wszelaki (2012) found that growers can increase marketable yield by using a high tunnel when compared with open field production. WMARS had a higher marketable weight than Spooner, which might be due to the higher temperatures and the longer season. This is in agreement with previous work we have done comparing high tunnel and open field management (Healy et al. 2016, Hodge et al. 2018). As we do not have fully replicated management systems in each location (there is only one high tunnel in Spooner and one at WMARS), the years serve as our replication of the systems within location and the lack of crossover interaction shows that the results of the tests of management system effects within locations are reliable. This shows that the caterpillar tunnel is a viable alternative to high tunnels at a lower cost for vegetable farmers.

In addition to their being a consistent effect of management systems within locations, there was a consistent performance of the different genotype between management and location. There were no significant changes in ranking of varieties across the three systems within a location, and there was only one significant change in rank across locations, with Defiant being the best variety in WMARS and Big Beef being the best variety in Spooner for marketable weight. Marketable count followed the same pattern as marketable weight, although as Defiant had smaller fruit it has a consistently higher marketable count. This shows that all varieties had higher performance in the tunnels and variety recommendations can be made for each location without having to make recommendations specific to management systems, this is in agreement with Jett and Read (2002) who found that almost every cultivar that performs well in the open field will perform better in the high tunnel. It is also in agreement with our previous results (Healy et al. 2016, Hodge et al. 2018). However, as there were system-specific differences in the prevalence of certain causes of unmarketability, such as blossom end rot, choosing varieties that are less susceptible to these particular problems for specific systems may be beneficial to growers. By using a representative group of tomato varieties, we hope our results are generalizable to other varieties that may be grown in tunnels in the NCR.

Between management systems, the open field had much higher levels of unmarketable yield compared to both tunnels. The open field is not protected from rain, causing more unmarketable fruit due to fluctuation in the soil moisture that can lead to fruit cracking and splitting (Peet and Willits 1995, Rogers and Wszelaki 2012). It also has higher disease pressure (Blomgren and Frisch 2007, Wells and Loy 1993), which is also a factor that may increase the unmarketable fruit proportion. Hornworms are more prevalent in the tunnels compared to the open field; this may be because they are protected from predation and for higher temperatures in the tunnels. Blossom end rot seems to be higher in the high tunnel compared to caterpillar tunnels and open field. This might be due to multiple factors such as soil nutrient imbalances, uneven moisture, higher temperatures or even faster vegetative growth (Rogers and Wszelaki 2012). This might be due to more moisture reaching the soil under the plants in the caterpillar tunnel from the edges, due to its narrower width, decreasing the chance of extremely low moisture levels in comparison with high tunnel.

 Disease

There were significant differences in disease incidence due to management. The open field had significantly higher TAUDPC followed by the high tunnel, then the caterpillar tunnel. The high tunnel may have had higher levels of humidity than the caterpillar tunnels, leading to higher levels of leaf mold in the high tunnel. This is in line with our previous trials that found significantly higher levels of TAUDPC in the open field compared to the high tunnel (Healy et al. 2016, Hodge et al. 2018), there were not caterpillar tunnels included in those studies. Where we could test the effects of location and year, these were also significant sources of variation, with Spooner showing lower TAUDPC than WMARS and with a trend towards higher TAUDPC over years. In the open field, the TAUDPC was mostly SLS+EB, while in the caterpillar tunnel and high tunnel it was mostly EB+LM. For PM there was no difference between management systems. This composition is important because comparing between the tunnels and open field, LM in the tunnels seems to trade off with SLS in the field. However, LM seems to progress much slower and affects fruit production less severely than SLS. In terms of yield reductions, it seems like SLS has a larger impact than LM in our study. While EB seemed less prevalent in the field, this may have been partly due to the preponderance of SLS which made scoring EB separately much more difficult in the field than in the tunnels. EB also showed more yearly variation in disease pressure than other diseases.

The open field is the least affected by LM, which was almost absent every year. This is consistent with the literature that show that leaf mold progresses faster in environments with high relative humidity and temperatures (Warren 2015). This type of environment is exactly what caterpillar tunnels and high tunnel provide. In addition, the high tunnel had higher temperatures, humidity and lower ventilation than the caterpillar tunnel which may explain the higher levels of LM present there. This is consistent across years and location. SLS is a disease that thrives when there is a prolonged leaf wetness and when there is splash when rain hits the soil or mulch, because the spore that is in the soil needs the moisture in the foliage to infect it. Therefore, with the open field being unprotected from rain, it has the ideal conditions for SLS to thrive.

Overall, the management system had a more pronounced effect on disease than the variety, but there were a few differences in varieites for particulary diseases and overall TAUDPC. In the high tunnel, Big Beef had significantly higher TAUDPC compared to all varieties except Pruden’s Purple. Damsel and Defiant performed significantly better than Pruden’s Purple for EBAUDPC. For LMAUDPC, Japanese Black Trifele and Paul Robeson were significantly lower than Big Beef in the high tunnel for years 2019 and 2020 where LMAUDPC levels were high enough to see significant differences.

Quality

Generally, both tunnels had higher ˚Brix levels compared to open field but there were no significant differences between tunnels. This may be explained by the higher temperatures in both tunnels compared to the open field and also that there were no significant differences in temperature between caterpillar tunnel and high tunnel. Another factor that might affect ˚Brix levels is that generally, both tunnels had lower soil moisture levels compared to the open field, increasing the ˚Brix levels (Aguyoh et al. 2015). The healthier foliage in the tunnels may also have been able to produce more sugars as fruit ripened due to higher sustained levels of photosynthesis. The higher levels at WMARS compared to Spooner can most likely be explained by the higher temperatures in that location (Gautier et al. 2008). Results show that CA levels increased in the open field compared to both tunnels, which confirms we saw in prior studies by Healy et al. (2016) and by Hodge et al. (2018). These studies also reported higher ˚Brix levels in the high tunnel compared to the open field. The tastings confirm what was found in the ˚Brix and CA, with higher perceived sweetness in the tunnels compared to the open field, and higher acidity in the open field compared to both tunnels, with no significant difference between tunnels. The high tunnel and caterpillar tunnel both showed a significantly higher overall flavor rating compared to the open field. This is most likely due to the higher sweetness/˚Brix and the lower acidity/CA levels. This goes in line with previous work from Healy et al. (2016) and by Hodge et al. (2018).

Significant differences between varieties were seen, supporting the strong effect on genotype on sugar accumulation. Damsel performed best across management systems, which may be due to the fact that this variety was bred specifically for flavor. Damsel also had generally higher levels of CA compared to other varieties, which may also be due to selection specifically for flavor. The results with Damsel were borne out in taste evaluations and farmer preferences for the variety. Damsel had the highest mean value for overall flavor although it was not significantly different than most other varieties in each system. Most varieties showed improved flavor in both tunnels compared to the open field, however, there was some variation in whether varieties had better overall flavor in the high tunnel or in the caterpillar tunnel.

On-farm trials

The qualitative results from on-farm trials align with our on-station trials, with the open field generally performing worse than both tunnels and no clear differences between the high tunnel and caterpillar tunnel. For the slicer market class, Damsel seemed to be the preferred variety due to its good performance and excellent flavor. The best variety in the open field on farm surprisingly was Paul Robeson for the heirlooms and Damsel for the slicers. For the tunnels, the best slicer was again Damsel and Japanese Black Trifele seemed to be the best heirloom.

Costs of production and revenue potential

The caterpillar tunnel with one row required the greatest number of person-hours of management, followed by the high tunnel, the estimated caterpillar tunnel with two rows and the lowest was the open field. The difference between tunnels and open field can be explained in large part because of the 13 harvests in the tunnels and higher yields versus only 6 harvests in the open field and the lower yields. Additionally, irrigation, ventilation, pruning, trellising and moving for rotation contribute to the differences in the labor requirements between all tunnels and open field.

Harvesting was the principal task requirement by person-hours between all management systems, with very similar amounts of time required for the caterpillar tunnel and high tunnel systems. Pruning and trellising person-hour requirements are very similar between caterpillar tunnels and high tunnel as same system was used in each management system. Second, the plants also had the same number of trellising and pruning work days. On the other hand, the open field used a Florida weave system that is a less intensive system in terms of person-hour for trellising and had fewer trellising workdays because of later planting and the different trellising system. This way of management also requires less intense pruning.

Differences in ventilation are evident in that both tunnels were either opened or closed 36 times during the 2020 season. The high tunnel requirements for ventilation were much lower due to its quick and efficient handle system used to open and close. On the other hand, opening a caterpillar tunnel is a completely manual procedure consisting of rolling the plastic inwards and the reverse for closing. This is a slow process requiring a moderate amount of effort.

The differences between caterpillar tunnels with one and two rows are notable when comparing the total hour requirements and thus it may be worth it to farmers to trial two rows of tomatoes in the caterpillar tunnels. However, two row caterpillar tunnels might have more disease pressure due to lower ventilation and also due to more rain reaching the plants because are closer to the edges of the tunnel or simple due to more competition for resources from the plants like water, light and nutrients. They might result in a decrease in marketable yield and overall plant health, making the potential profit smaller. More studies need to be done in the future regarding this.

There are also differences between systems in terms of initial costs. This can be divided into the cost of purchase and the cost of labor to build each system. The purchase of a high tunnel is the most expensive at around $13,000 compared to $1,500 for a caterpillar tunnel. The cost of a caterpillar tunnel appears to be much lower, but a high tunnel has enough area for about 300 plants compared to about 55 plants in a single row in a caterpillar tunnel. Therefore, if one row caterpillar tunnels are used, six of them are needed in order to grow the same number of plants as a high tunnel, and three two rows caterpillar tunnels are needed to hold about 300 plants. This bring the cost of purchase to about $9,000 for a single row caterpillar tunnel and $4,500 for a double row, which is still considerably lower than the $13,000 cost of a high tunnel. Some of the reasons that caterpillar tunnels have lower costs is the cheaper materials used. Caterpillar tunnels are made mostly out of PVC bows, plastic film, rope, and anchors. On the other hand, high tunnels use large amounts of galvanized steel, wood boards, screws, bolts, wiggle wires, plastic films, and anchors, among many others raising the cost of materials.

The costs of building the different management systems also vary greatly. As expected, the higher complexity of high tunnels is translated into more hours of construction. A high tunnel needs about 400 hours of labor to be constructed due to the multi-step process. Some of these steps, such as installation of the plastic film or assembling the frame, require at least four people at a time to complete the task. Also, some of these tasks might require tractors, trucks and other equipment to make the job more efficient. Furthermore, the end walls and side-walls need to be installed as well as ensuring functional opening and closing mechanisms. The installation of a caterpillar tunnel is significantly easier and faster. A crew of two people can install a caterpillar tunnel in a morning or eight person hours, totaling 48 person hours for the single row systems and 24 person hours for the double row caterpillar tunnel.

A cumulative net revenue projection for the first 10 years is possible given the purchase cost, building costs and yearly management. The results reveal initial costs being much higher in all tunnels, but due to the significantly higher marketable yield in these systems, the revenue that tunnels bring is enough to increase the net income compared to the open field. At year one, the systems with negative net income are the high tunnel and one row caterpillar tunnel due to their high initial purchase and the high tunnel’s considerable building cost. However, these systems recover rapidly, with the one row caterpillar tunnel achieving a positive cumulative net income at year 2 and the high tunnel attaining positive net income at year 3. The open field falls behind the other management systems for a different reason; the factor most prominent in the lower yearly net income is the low productivity, therefore bringing low revenue to the farm.

It is encouraging to see that caterpillar tunnels are a financially viable alternative that require less initial investment compared to the considerable investment that high tunnels bring. This sizable investment can be a challenge for small vegetable farmers who may experience difficulty accessing credit. Another factor to consider is that the investment for the high tunnel needs to be done all in year one, but caterpillar tunnels can be purchased one at a time. Also, there is a higher risk for the farmer with the high tunnel in the case of destruction of the tunnel by a storm or a crop failure. This risk can be partially mitigated by the smaller investment of caterpillar tunnels.

2. Selecting varieties for high tunnel or caterpillar production with excellent flavor and a sustained high fruit set early in the season and varieties with disease resistance, flavor and dual purpose use under field conditions.

At the end of the F5 generation, we have the lines listed in Table 6 remaining. In the end we were left with progeny from 7 of the original parents, with other families dropping out due to poor performance. The parents remaining in the crosses evaluated are described below, and the quality and production results in the high tunnel of the lines in Table 6 are presented in Table 7.

Defiant: F1 hybrid that has high resistance to late blight (resistance genes Ph-2 and Ph-3) and intermediate resistance to early blight. It also has high resistance to Fusarium wilt races 1, 2, late blight, and Verticillium wilt (Johnny’s Selected Seed, 2021) 


OSA404: a cross between WI 55 and a disease-resistant North Carolina State inbred, selected by the Organic Seed Alliance for disease tolerance and flavor over several years. Received in 2014 as an advanced line and maintained by selfing. 


A6: an Amish heirloom selected for Midwest adaptation by Craig Grau, a retired plant pathologist at UW Madison. 


Japanese Black Trifele (JBT): crack-resistance heirloom with a smoky flavor maintained by Ken Greene at the Hudson Valley Seed Company 


Crimson Sprinter: an heirloom from Ontario, CA with partial Septoria leaf spot resistance, earliness and good flavor. 


Summer Sunrise: a cross between ‘Golden Dwarf Champion’ and ‘Green Giant’ in 2005 by Patrina Nuske Small. Released from the "Dwarf Tomato Project" in 2012 as a selection made by David, Susan, Neil Lockhart and Justin Morse, with additional help from Ted Maiden and Craig LeHoullier. Tart, sweet, intense flavor. 


P321: breeding line from Jim Myers at Oregon State University as a cross between Indigo Rose and Ananas Noir, selected for larger size and excellent flavor. Contains the anthocyanin trait from Indigo Rose giving purple shoulders.

Production traits

Marketable Weight

Breeding line had an effect on marketable weight (p-value < 0.001). Overall, the check varieties had a higher marketable yield than the breeding lines. There was strong evidence for interaction between variety and management (p-value < 0.001). The nature of the interaction was scalar. For the high tunnel, Caiman was the variety with the highest marketable yield, followed by the rest of the check varieties (Table 4). The best performing breeding line was CSDE-F6-46. For the open field management, Defiant had the highest marketable yield, followed by the JBDE-F5-28 breeding line. The highest yielding breeding lines in the high tunnel were, in decreasing order, CSDE-F6-46 (7.4 kg/plant), O4DE-F5-44 (7.1 kg/plant), and O4DE-F5-43 (7.0 kg/plant). The lowest yielding breeding lines in the high tunnel were, in increasing order, O4JB-F5-MV1-115 (1.6 kg/plant), P3SS-F4-61 (3.2 kg/plant), and O4JB-F5-MV1-116 (3.5 kg/plant). Interestingly, the top three varieties are descendants of Defiant, and two of the crosses are with OSA404. At the same time, two of the lowest yielding varieties are descendants of OSA404.

Marketable Count

Breeding line also had an effect on marketable count (p-value < 0.001). Overall, the check varieties had a higher marketable count than the breeding lines. There was strong evidence for interaction between variety and management (p-value < 0.001). However, the nature of the interaction was primarily scalar rather than crossover. The breeding lines with the highest marketable count in the high tunnel, in descending order, were JBDE-F5-31 (61 fruits/plant), and Defiant descendants CSDE-F6-46 (53 fruits/plant), and JBDE-F5-28 (49 fruits/plant). The breeding lines with the lowest marketable count in the high tunnel were O4JB-F5-MV1-115 (4 fruits/plant), O4JB-F5-MV1-116 (15 fruits/plant), and A6JB-F5-34 (15 fruits/plant).

Average weight

Breeding line had a significant effect on average weight (p-value < 0.001). There was strong evidence for interaction between variety and management (p-value <0.05). Average fruit weight ranged from 87g/fruit (JBDE-F5-31) to 365g/fruit (O4DE-F5-43). Both are descendants of Defiant, the first being a cross with the smaller-fruited Japanese Black Trifele, and the second with larger-fruited OSA 404. The breeding lines with the highest average fruit weight in the high tunnel were O4DE-F5-44 (381 g/fruit), O4DE-F5-43 (372 g/fruit), and O4JB-F6-5 (321 g/fruit). The breeding lines with the lowest fruit weight in the high tunnel were JBDE-F5-31 (87 g/fruit), P3SS-F4-61 (99 g/fruit), and JBDE-F5-28 (110 g/fruit).

Proportion unmarketable

The effect of breeding line on the proportion of unmarketable weight was significant (p-value <0.001). There was strong evidence of interaction between management and genotype for the unmarketable proportion (p-value <0.001). For the high tunnel, stem side cracking was the main reason for the primary cause of fruit unmarketability (47%), followed by radial splitting (16%), and hornworm damage (6%). Similarly for the field, stem side cracking accounted for 64% of the counts, followed by radial splitting (10%), and rodent damage (10%). Blossom end rot accounted for 1% of the unmarketable causes in the field, and 5% in the high tunnel. In the high tunnel, Defiant had the lowest proportion of unmarketable weight, followed by the breeding lines JBDE-F5-28, JBDE-F5-31, CSDE-F6-46, and CSDE-F6-47. The first two are a cross of Defiant with JBT and the last two a cross of Defiant with Crimson Sprinter. The primary causes of unmarketability in the high tunnel were blossom end rot (28%) and interior blemish (28%) for JBDE-F5-28, side stem crack (36%) and insect (27%) for JBDE-F5-31, stem crack (38%) and insect (19%) for CSDE-F6-46, and radial splitting (32%) and side stem crack (16%) for CSDE-F6-47.

Disease

In general we did not see substantial differences between breeding lines for disease incidence. This is likely due to extreme pressure in the field and low pressure in the high tunnel. For specific diseases, we did see some differentiation among lines. The lowest EBAUDPC scores were Defiant in the field and Paul Robeson in the high tunnel. Defiant's low Early blight score in the field is consistent with its history of breeding for intermediate EB resistance and high late blight resistance (Johnny’s Selected Seed, 2021). The breeding lines with the lowest EB scores in the high tunnel were JBDE-F5-28 O4DE-F5-43, and O4DE-F5-44. For the field, O4JB-F6-5, P3SS-F4-61, and JBDE-F5-31 had the lowest scores. The breeding lines with the lowest leaf mold score in the high tunnel were O4DE-F5-43, O4DE-F5-44, and JBDE F5-31. The severity of leaf mold was significantly higher in the high tunnel than in the field throughout the season.

Table 8 shows the means for each breeding family for each production and disease trait of the plants grown under the high tunnel management system. A6JB corresponds to the A6 by Japanese Black Trifele cross, CSDE to Crimson Sprinter by Defiant, JBDE to Japanese Black Trifele by Defiant, O4DE to OSA 404 by Defiant, O4JB to OSA 4040 by Japanese Black Trifele, and P3SS to P321 by Summer Sunrise.

Fruit quality

°Brix and Citric acid

In the fruit quality analysis, both °Brix and Citric Acid (%) showed significant evidence for the genotype effect (p-value <0.05). The breeding line P3SS-F4-61 had the highest °Brix value (7.15), and the highest CA% (0.51). Pruden’s Purple had the lowest °Brix value (4.13) and A6JB-F5-35 had the lowest CA% (0.23).

Sensory Analysis

Within the breeding lines, P3SS-F4-61 and O4JB-F5-MV1-115 stood out for particular attributes, and for the overall evaluation, P3SS-F4-61 had the highest score, followed by Damsel, and CSDE-F6-47 (Table 9)

Correlation among traits

Table 10 shows the phenotypic correlations among 4 production traits and 5 disease traits for the breeding lines. Marketable weight shows a strong positive correlation with the marketable count, resistance to early blight and proportion unmarketable.

Table 11 shows the phenotypic correlations between the fruit quality traits and marketable weight. Marketable weight shows a high negative correlation with fruit acidity, umami, and intensity. Titratable acidity is positively correlated with fruit acidity (tasting), and °Brix is also highly positively correlated with the sweetness perceived by the tasters. CA and °Brix have a high positive correlation (0.87). CA is also positively correlated to Texture, Sweetness, Acidity, Umami, Intensity, and Overall. Marketable weight has a negative correlation with all the flavor components.

On-farm evaluation

Participating farms received 6 breeding lines from different families to try in their farming systems. All the varieties evaluated on-farm were grown in high tunnels. At the end of the season, each farmer submitted a qualitative evaluation form that indicated which varieties they preferred with respect to productivity and flavor along with any other relevant observations. Table 12 presents a summary of all the evaluations received from 2019 and 2020. Is important to note that not all participants received the same breeding families, and overall, the summary includes the evaluations of 2 or 3 farms per cross.

The CSDE family had high productivity with good flavor, for some farmers it was the “winner” in terms of flavor compared to the rest of the lines. One farmer mentioned that ripening started late compared to their commercial varieties but did not specify how late (days or weeks). O4JB had high yields and great flavor and was categorized as the best overall. In terms of earliness, it was between Damsel (earliest) and Caiman (a little later than Damsel). JBDE had great flavor, and productivity varied from medium to high. A6JB was identified as a typical heirloom with a very sweet and good flavor. It had low disease resistance and low productivity. O4A6 had medium disease resistance, good flavor, and productivity ranged from low to high. Farmers also indicated that higher amounts of radial splitting and cracking, and it must be sold soon after harvesting due to its soft skin.

Discussion

Performance of field selections was disappointing in terms of disease resistance, and we do not expect to release varieties from this particular set of crosses, but we are using the populations created to study field disease resistance as part of two successful grant applications which were funded. It is likely that with the severe septoria and early blight pressure in the midwest, we will need to incorporate stronger qualitative resistance alleles in addition to the quantitative resistance we were hoping to select for. We are continuing this work as part of an OREI-funded project led by Purdue University and a Wisconsin Organic Initiative grant that will fund a PhD student.

In the high tunnel results are much more promising. The best lines are nearly stable and will be tested for on-farm performance as part of our Wisconsin Organic Initiative grant and a different OREI-funded project led by Oregon State University. In addition, we will be working with independent plant breeders in the Upper Midwest to build on this work as part of a third OREI-funded project that we are leading in collaboration with SeedLinked and Organic Seed Alliance.

Promising breeding lines

Based on the results of our evaluations for production, disease, and fruit quality traits, we have chosen several breeding lines for further field evaluation.

CSDE-F6-46 had the highest marketable weight (7.4 kg/plant) of all the breeding lines. The average weight of the medium-sized, round fruit was 139g. It had a low proportion of unmarketable weight (0.1), and the primary cause of unmarketability was stem side crack. Its early blight AUDPC was close to the overall mean of only the breeding lines in the high tunnel (791 and 751 respectively). It was not particularly affected by any of the rest of the diseases evaluated. In terms of the sensory evaluation, it had a low Overall score (2.6), low Intensity (2.4), and high Appearance (4.1).

In the same breeding family, the line CSDE-F6-46 had a lower marketable weight (6.3 kg/plant), a slightly higher unmarketable proportional weight (0.18). It had a similar Total disease AUDPC to CSDE-F6-46 (no significant difference). On the other hand, this variety did well in the sensory evaluation. It had the second-best score in the Overall category (3.4), a high intensity (3.4), and high Acidity and Sweetness (3.2 both). It also had high CA (%) (0.45), and intermediate °Brix (5.4). This line could be further selected, considering that the expected response to selection increases yields to 1.45 kg/plant keeping close attention to the flavor evolution, since yield and flavor are negatively correlated. It is also important to note that farmers liked this family and found it to be productive with good flavor.

JBDE-F5-31 and JBDE-F5-28 both had intermediate marketable weight (5.3 kg/plant for both, no significant difference). The average fruit weight was low (87 and 108 g/fruit respectively), resulting in a medium-sized fruit. Both had a low unmarketable proportion (0.08 and 0.09), with the primary cause for unmarketability being stem side cracking, which is common in Japanese Black Trifele, one of the parental lines. In terms of flavor, JBDE-F5-28 did better Overall (3.2), with high Intensity (3.1), Sweetness (3.1), and low Acidity (2.4). This family was liked by farmers because of the medium to high yields, the sweet flavor, and the overall good health of the plants. However, testers commented that the skin of both lines was too thick.

P3SS-F4-61 had the best Overall flavor score, with high acidity, sweetness, intensity, and umami. In terms of production, it had a low marketable weight in the high tunnel (3.1 kg/plant) compared to the rest of the slicer breeding lines (overall mean 5.3 kg/plant). The marketable weight variance was high, and this trait also had a high heritability, so there is room for yield increase when selecting future generations. Average fruit weight had a high variability within plants, with a confidence interval from 58 to 140 g. Because this is an F4, uniformity has not yet been achieved, and that has been made clear with fruit size variation. The proportion of unmarketable weight was high (0.58), and the primary cause of unmarketability was radial splitting (85%). Better irrigation management at the time of fruit ripening could help decrease the amount of unmarketable fruit, and at the same time increase the marketable weight. In terms of disease, it had scores close to the mean in all the diseases evaluated and did not show high susceptibility or resistance to any of them.