Cover Crops Under Cover: Evaluating Costs, Benefits, and Ecosystem Services of Cover Crops in Year-Round High Tunnel Production Systems

Please note, due to a relatively large amount of data generated from the project we have summarized some key results and discussion points in the paragraphs below.  Once published citations for each objective become available, we will update the final report.  These works will contain more complete results and analyses than those presented here.  The project team would like to acknowledge the analysis and writing assistance from supporting staff, including Mr. Alex Butler and Mr. Matthew Ernst (Objective 1) and Ms. Victoria Stanton (Objective 3).

Objective 1: Results from the high tunnel producer survey. 

Demographics of respondents.  Nearly three-quarters (73%) of responding producers operated high tunnels in Kentucky, with another 11% from Georgia, Tennessee and Alabama, collectively.  Males accounted for 59 percent of respondents, and females 41 percent.  One-fourth of respondents were in the 35 to 44-year age range, and 19 percent were under 35; only one operator reported being less than 25 years old. More than one-third (37 percent) of operators were 55 or older.  Growers were generally experienced with high tunnels, with half of the respondents indicating they have used high tunnels for 10 or more seasons. 

Sales and Marketing Channels.  Respondents were asked to estimate the percentage of their produce sales originating in the high tunnel. The average response was 46 percent of produce sold from the farm is grown in the high tunnel(s). There were 96 of 106 respondents that indicated their use of produce marketing channels before and after installing their high tunnels. Their responses provide insight into both the primary marketing channels of the responding producers and the expected impact of high tunnels upon their farm sales.  The data indicate that producers tend to diversify market channels after installing high tunnels. The rates of respondents using community supported agriculture (CSA) and community farmers markets remained about the same after installing the high tunnels. Other market channels increased after high tunnel installation. The largest increase, by percentage of use, came in those selling produce at an on-farm retail market. This may indicate producers preferring to pursue a market channel with fewer transaction costs, like transportation and packing.

High tunnels helped respondents extend their produce growing and marketing season while diversifying the number of products offered. More than 90 percent of respondents said early- and late-season market activity “increased a lot” or “increased some” after installing their high tunnel. High tunnels also diversified product offerings, with 45 percent of respondents saying their variety of products marketed increased “some” after high tunnel installation and an additional 20 percent saying the variety increased “a lot.”

Those installing high tunnels also increased marketing to wholesale channels: groceries, restaurants, schools and other institutions, and auctions. The increase in serving schools and institutions is noteworthy, as an additional 10 percent of producers sold through those channels after installing their high tunnels. Producers were very optimistic about the school/institution market, with 94 percent of producers selling to schools expecting those sales volumes to increase.  Interestingly, those using high tunnels also indicated optimism about the demand for local food in their area, with 24 percent expecting local food demand to increase significantly and 49 percent expecting demand to increase somewhat. Less than 10 percent said they expected any sort of decrease or had no good basis for knowing.

High tunnel production practices.  Survey respondents indicated a mix of conventional and organic growing practices in their high tunnels. One-fifth said they used primarily conventional growing techniques, with an additional one-third incorporating some organic along with conventional methods. The most common response (36 percent) was that high tunnel production was according to organic standards but not certified. Ten percent of respondents indicated their high tunnels were certified organic.  More than half of high tunnel operators said they used high tunnels for year around production always or sometimes. 

Testing soil, rotating crops and scouting are the most regular production practices reported by high tunnel operators. Cover cropping, followed by solarization and thermal banking, were the next most frequent practices reported. Rainwater catchment systems were less frequent overall, but those using rainwater catchment were more likely to report regular use of rainwater catchment as a practice. Biofumigation was ranked as the least likely management practice by those surveyed.

High tunnel production challenges. Weed, insect and disease management were ranked as the most challenging production situations among a list of eight common management challenges (Figure 1). Soil fertility and space management were next most challenging, as evidenced by responses of “serious problem and difficult to manage” or “regular problem but readily managed.”  Along with weed management, labor availability and salinity were most frequently ranked as serious problems difficult to manage. The least challenging area among this list was oversupply for current markets.

Areas of Biggest Questions.  When asked to rank areas in which they had questions, more high tunnel operators said they were “in good shape” for crop selection than any other area. Cover crops were most likely to be rated as “not applicable” because of a significant number of high tunnel operators not using cover crops.  High tunnel operators were slightly more likely to say they “need lots of help” in the areas of marketing, weed management, and soil fertility management. Insect pest management and disease management were most likely to be ranked as “making some progress.”

More high tunnel operators said peer-farm high tunnel demonstrations were “extremely helpful” than any other information source. Cooperative Extension programs were indicated as the resource most frequently used, also receiving favorable ratings. High tunnel operators also showed a penchant for accessing online resources and ranked online publications and YouTube or online videos to be useful. NRCS programs received high marks for being “extremely helpful,” likely reflecting resources available through cost share programs in Kentucky and other states. Other agency programs and trade association workshops were the resources respondents were least likely to draw upon for information about high tunnel production (Figure 2).

Objective 2: Results from the ecosystem services of cover crops in high tunnels experiment. 

Climate Data.   Temperatures in winter of 2018 were more extreme in 2018 than in 2017, with lower winter temperatures and higher early summer temperatures.  The cooler winter temperatures led to less high tunnel cover crop growth at both sites in 2018, discussed below.  Interestingly, our climate data show that minimum temperatures in high tunnels can be lower than the outside temperature, though with sun, they warm rapidly.  This is consistent with other literature documenting high tunnel temperature dynamics, but it is somewhat counter to the conventional wisdom that high tunnels offer consistent cold protection above field temperatures. 

Cover Crop Performance and Composition.  In both Kentucky and Tennessee, total cover crop biomass was similar in all treatments, though more biomass was produced in 2016-17 compared to 2017-18.  When averaged across all cover crop treatments, in both study years the total cover crop biomass was at least twice as great in Kentucky as Tennessee (2017: KY = 5.3 tons/ac, TN = 1.9 tons/ac; 2018: KY=3.8 tons/ac, TN = 1.4 tons/ac). 

In Kentucky, weeds contributed a large component of the cover crop treatment biomass.  There was greater weed biomass (p = 0.004) in the clover cover crop treatment, compared to the wheat and clover/wheat mixture (Figure 3).  Wheat biomass in Kentucky was greater when winter wheat was grown alone compared to grown in with clover, likely due to the lower wheat seeding rate in the mixture.  When averaged across both years, the wheat alone treatment had the greatest nitrogen content (141.6 lbs N/ac, p=0.0483) compared to the clover alone (115.6 lbs N/ac) or the clover and wheat mixture (115.1 lbs N/ac), although the means were not different from one another.  

Overall, the clover was not a very effective weed competitor at the Kentucky site.  This site is part of the “transition zone,” and falls within the southernmost region of many northern weed species, and the northernmost region of many southern weeds.  Further, the site is located on fertile, prime agricultural soils in the state, and in nutrient-rich high tunnel soils.  As such, legume cover crop species grown alone may not be the ideal choice for such soils, as their nitrogen fixing services may be outweighed by their lack of weed suppression.

The Tennessee site did not have the weed pressure seen in Kentucky, and the clover performed well relative to weeds.  This is perhaps due to the relatively lower fertility, clay soils at the Tennessee site, in which legume cover crops can be effective competitors in low weed pressure situations.  The clover biomass was greater when grown alone than when grown in mixture in both years (Figure 3).  When grown alone, the wheat biomass was similar between the two years, although the wheat biomass in the mixture was greater in 2017 compared to 2018.  Weed biomass in Tennessee was not affected by treatment, year, nor the interaction of the two.  In Tennessee cover crop N content was affected by treatment (p=0.0189), with the clover alone having a higher N content (58.1 lbs N/ac) versus the wheat alone (43.7 lbs N/ac), and the clover + wheat (53.2 lbs N/ac), which did not differ significantly from either species grown alone.  

Soil Quality. In Kentucky, we saw no significant effects of cover crop treatment on soil parameters, either through main effects or interactions.  Nearly all soil parameters, except phosphorus and zinc, varied by depth, which is expected due to the stratification of nutrients through soil layers.  All soil parameters, except for zinc, changed significantly over the course of the experiment.   For reasons that need further explanation, many soil parameters had levels that increased from year 1 to year 2, then decreased from year 2 to year 3.  This pattern occurred in many of the soil parameters, and was a similar trend in the significant date by depth interactions observed for soil P, Zn, soluble salt, and total carbon and nitrogen levels (data not shown).  However, these changes did not vary by treatment, and may have been a function of regional climate patterns or variability due to production practices, as similar trends were seen in Tennessee (discussed below). 

In Tennessee we observed cover crop treatment effects on phosphorus, pH and soluble salt levels.  Phosphorus were greater levels were greater in the continuously cropped (62 lbs/ac) and clover+wheat (53 lbs/ac) treatment, though the wheat and clover alone (49 lbs/ac each) treatments were not significantly less than the clover+wheat mixture.  Potassium and magnesium and zinc levels all declined within each treatment throughout the course of the experiment, but did not vary between treatments within each year.  As such, no one treatment significantly buffered or drove the fertility declines seen in the experiment.  

 Several soil parameters (phosphorus, calcium, magnesium, zinc, soluble salts, and total soil carbon and nitrogen) had significant date by depth interactions.  The trends for phosphorus and calcium were similar, with values varying by depth, with lower values in the deeper layers and higher values in the shallow layers throughout the experiment.  However, within each layer, levels did not differ from starting levels by the end of the experiment.  Magnesium levels varied by depth and did not differ at either depth between year 1 and year 2; however, in year 3 Mg levels dropped and did not vary by depth.  Zinc levels also varied by depth throughout the experiment, and declined from year 1 to year 2, but did not change from year 2 to year 3.   Soluble salts varied by depth throughout.  They did not differ significantly from start to finish in the surface layer, but were reduced by year 3 in the deeper soil layer.  However, none of these levels are problematic for tomato or lettuce production.  Similarly, total soil carbon did not differ in the surface layer from start to finish, but was slightly reduced in the lower soil sampling depth by the end of the experiment.  Soil pH varied by depth in 2016, but by the end of the experiment did not differ significantly within or between soil sampling depths.  pH in the soil surface layer dropped from 5.9 – 5.6 during the three-year study, but was not unexpected due to the lack of application of lime or other pH altering amendments during the experiment. 

Of particular note to soil quality issues in high tunnels, we observed a decline in soluble salts throughout the experiment at both sites.  Soluble salts can be production-limiting to certain crops, and are prone to building up in high tunnels due to lack of flushing rains under the plastic cover.  Soil K levels also decreased significantly every year at each sampling depth, concomitant with the decline in soluble salts. This decline in K was likely the source of the tomato ripening disorder (yellow shoulder) observed in the final year of the experiment at both sites.   

Nitrate Leaching.  We did not see significant treatment or seasonal effects on nitrate leaching losses in the ion exchange resin lysimeters.  Overall, leaching losses were quite low, and were under 5 lbs NO3/ac at each sampling date.  This could reflect very little free water percolating through the soil depths where the lysimeters were located, which was largely below the rooting zone (20 inches).  This is probable, as soil conditions were fairly dry at each excavation/deployment.  This would indicate that as a conservation practice, high tunnels effectively reduce nitrate leaching in soils under the structures, irrespective of the crops and cover crops used in this experiment.  These low levels may also be due, in small part, to incomplete de-sorbtion of nitrate ions from the ion exchange resin used in the resin lysimeters.  The resin inside the lysimeters was only extracted with a single round of KCl solvent, and additional ions may have been released with additional rinses.

Tomato Yield.  Marketable and total tomato yields (by weight) did not differ among treatments in Kentucky or Tennessee in either year, with the exception of total yield in Tennessee in 2017 which was lowest for the wheat treatment and similar for the other three treatments (data not shown).  Total marketable yields were higher in Kentucky than in Tennessee when averaged across both years (p<0.0001, Kentucky = 64.2 lbs/100 ft², Tennessee =  39.3 lbs/100 ft²), and yields across both sites were greater in 2017 than in 2018 (p<0.0001, 2017 = 80.9 lbs/100 ft², 2018 = 23.8 lbs/100 ft²).  High   levels of the tomato ripening disorder yellow shoulder (YSD) were observed at both sites, especially in 2018 (data not shown).  Prevalence of YSD is related to soil nutrient imbalances, and is common in high tunnel tomato production.  We discuss the relationship between yield and soil quality in the section below. 

Correlating Relationships Between Cover Crops, Yields, and Soil Quality.  Although we did not see cover crop treatment effects on tomato yield or soil quality through analysis of variance, we did find correlations between these attributes.  Tomato yields were strongly positively correlated with total cover crop biomass (r_s = 0.71, p<0.0001) and the wheat cover crop biomass (r_s = 0.63, p=0.0027) in Kentucky, and moderately correlated with these variable (r_s = 0.43, p=0.0089; r_s = 0.46, p=0.0246, respectively) at Tennessee.  In Kentucky yields were also moderately correlated with the nitrogen content in the cover crop (r_s = 0.56, p=0.0012).  These data indicate that there is a positive relationship between cover crop biomass and yield, and further, that yields are not positively or strongly correlated with legume cover crop biomass. 

The effect of low soil potassium levels on yields in the second year of the experiment has been discussed above.  However, the correlation data suggest a slightly more nuanced story.  In Kentucky, yields were weakly positively correlated with soil K content (r_s = 0.36, p=0.0223 in the 0-6” layer, (r_s = 0.36, p= 0.0244 in the 6-12” layer) and were moderately positively correlated in Tennessee (r_s = 0.54, p=0.0223 in the 0-6” layer, r_s = 0.55, p<0.0001 in the 6-12” layer).  Yields were moderately negatively correlated with magnesium content in Kentucky (r_s = -0.59, p<0.0001 in the 0-6” layer, r_s = 0.49, p=0.0014 in the 6-12” layer) and very weakly negatively correlated in Tennessee (r_s = -0.15, p=0.3253 in the 0-6” layer, r_s = -0.14, p=0.3395 in the 6-12” layer).  However, in both locations yield was more strongly correlated with the “Hartz ratio,” a measurement of the ratio of exchangeable potassium and magnesium content (millequivalents of soil exchangeable K/√Mg).  Yields were strongly correlated with Hartz ratio in Kentucky (r_s = 63.15, p<0.0001 in the 0-6” layer, r_s = 0.64, p<0.0001 in the 6-12” layer), and moderately correlated in Tennessee (r_s = 0.57, p<0.0001 in the 0-6” layer, r_s = 0.56, p<0.0001 in the 6-12” layer).  Finally, in Tennessee, yields were also moderately positively correlated with active carbon (r_s = 0.56, p<0.0001 in the 0-6” layer, r_s = 0.51, p=0.0002 in the 6-12” layer), indicating a positive relationship between labile carbon and yields on these heavy clay soils. 

As such, these data indicate that the declining yields observed between the two years are more than simply a function of low soil K levels.  In Kentucky, they indicate a fertilizer salt imbalance rather than an overall decline in fertility may be more of the driving factor in yield and quality reductions.  In Tennessee, yields were moderately correlated with K content, Hartz ratio and active carbon (POXC).  This indicates that yield increases are correlated with increasing soil carbon content and increasing K fertility on these heavily-weathered clay soils. 

Economic Trade-Offs of Winter Crops and Cover Crops.  In Tennessee and Kentucky, two lettuce crops were harvested in the 2016-2017 winter season, and one crop in 2017-2018.  Gross profits ranged from $2.06 - $2.99 per square foot for each harvest, but harvest quality and yields were variable from year-to-year and between sites.  For example, almost half of the first lettuce harvest in Tennessee was unmarketable, largely due to bolting and tip burn (data not shown).  In Kentucky, the second harvests in both years were significantly less than the first harvests, and of lower quality.  Thus, although cover crops do come with both the direct costs of the cover crop practice, they also come with opportunity costs associated with forgoing a winter cash crop.  However, the profit and predictability of the winter cash crop depends upon weather and viable direct markets in winter. 

Our results to do not indicate a strong effect of specific cover crop direct ecosystem service benefits that would come with input cost reductions or other production cost savings.  However, correlation data do indicate that cover crops in general, and nutrient scavenging cover crops in particular, may confer yield and other soil health benefits in high tunnels.  Additional research is needed to improve our understanding of nutrient cycling processes when utilizing legume versus non-legume cover crops in high tunnels, as these unique environments are often more nutrient-rich and subject to different factors driving decomposition (e.g. temperature and moisture conditions) than the open field.  However, the results of this experiment do indicate that cover crops may play a different and unique role in high tunnel crop rotations and that cool seasons cover crops do not negatively impact warm season crop yields. 

Objective 3.  Results from the novel cover crop for high tunnel evaluation. 

Evaluating Cover Crop Accessions.  Cover crop evaluated included accessions that are known to perform well in the southern region in the field, as well as some new accessions that may confer benefits to high tunnel management for our region.  At the time of this writing, biomass data for the weed-free cover crop subplots have been analyzed, and are presented in Figures 4 and 5.  The majority of the warm season cover crops we evaluated were considered viable high tunnel cover crops for the Kentucky site where this evaluation occurred, with the exception of the brassica cover crops, which are discussed below.  Cover crop biomass was analyzed within functional group (brassica, legume, grass, or other), with functional groups compared to one another. 

Warm season grasses and the sesame cover crops had the greatest biomass each year, but as a group, grass biomass did not differ from summer legumes in 2017.  In summer 2018, sesame did not differ from summer legumes.  Biomass in 2018 was less than in 2017 in all accessions, likely due warm summer weather which made germination challenging.  Within the legume group, the sunn hemp varieties had the greatest biomass, although differences were not significant in 2018.  The varieties of sunn hemp evaluated included those with reported nematocidal effects (AU golden), as well more widely available varieties.  We found all sunn hemp varieties to be fast growing and viable for high tunnels.  However, the fibrous nature of the stalks as well as the quantity of biomass could make termination by mowing challenging if a producer is not mindful of the limits of their equipment.  The two varieties of cowpea were selected to provide producers management options with forage peas that are popular field cover crops in the region. Iron clay cowpea is a reliable summer cover crop throughout the region and is reported to have some nematocidal effects.  Chinese red cowpea is a lesser known forage crop, but is later maturing and has a more vining growth habit that ‘iron clay.’  Both cowpea varieties produced similar amounts of biomass each year with similar management requirements, and decomposed readily when incorporated (Figure 4). 

The grasses selected included two millet crops (German and Japanese), which were fast growing, high biomass cover crops.  The German millet was later to mature than Japanese millet, offering some options for cover crop rotation length before these covers set prolific seed at maturity.  Teosinte, a lesser known forage crop more widely recognized as the wild progenitor of modern-day corn, was also evaluated.  It consistently produced great amounts of biomass and could be used as an alternative to sorghum-sudangrass.  Sesame was also evaluated, as it is known to be a non-host to pest nematodes and as a member of the Pedaliaceae family, would also offer some variation in botanical family in crop rotations, as no commercial vegetable crops are closely related.  Although sesame was slower in growth than some of the accessions, it produced a weed suppressing canopy and the copious biomass was easily managed with mowing even at maturity, offering a high biomass and novel alternative summer cover crop. 

Winter grass cover crops consistently produced the greatest quantity of biomass, though in 2018-2019 biomass did not differ between grass and brassica cover crops.  Within the grasses evaluated, cereal rye consistently produced the greatest biomass, though it did not differ significantly from triticale in 2017-2018.  All of the grasses were deemed viable as they reliably overwintered, with the exception of festulolium, which experienced significant tip burn and stunting in winter 2017-2018. Further, as a very fine-structured grass, it was a weak competitor with winter annual weeds (data not shown).  In general, we found that grass cover crops that performed well in the field in Kentucky to be viable as cool season high tunnel cover crops.  The legumes selected represented a variety of less winter-hardy alternatives to the medium red clover typically grown in as a field cover crop in the Upper south.  The clovers demonstrated great variability year-to-year, and experienced significant winter kill in winter 2017-2018 .   

Our participating growers were specifically interested in using cover crops to suppress plant parasitic nematodes.  Certain brassica cover crops have been used as biofumigants for nematode suppression through naturally occurring isothiocyanates in the cover crops.  These crops are typically grown in the cool season, but it is known that these compounds aiding in plant parasitic nematode suppression are produced in higher quantity under warm conditions.  As such, we evaluated brassica cover crops as both warm and cool season cover crops. Although growth was much faster in the warm season, low germination due to high soil temperatures as well as flea beetle insect pest pressure limited the viability of the brassica covers evaluated during the warm season.  The cool season brassicas evaluated had much greater biomass and limited pest issues compared to those grown in the summer, and performed particularly well during the milder winter 2018-2019 season (Figure 5). 

In general, cool season cover crops required much of the winter and early spring to produce appreciable biomass. For the accessions evaluated in this trial, this would require the producer to forgo a cool season cash crop.  In the upper south when winter conditions may limit winter production, this may be an option.  However, for much of our region high tunnel production occurs throughout the winter months, and thus summer cover crops may be the most economically viable option for producers.  Summer covers offer their own challenges, but most obtained appreciable biomass in 30-45 days during mid-summer. 

At the time of this writing, all plant material data have been collected and statistical analyses are underway.  Additional data on cover crop weed suppression is currently being analyzed, including weed community dynamics in each accession as well as cover crop biomass and quality with unchecked weed pressure.  Cover crop quality data, including carbon and nitrogen content and potentially available nitrogen from NIRS protein analysis are also being analyzed.  We have one peer-reviewed manuscript in preparation for this work, as well extension/outreach deliverables, described below.   

Developing Growth Models for High Tunnel Cover Crops.  We have completed some initial modeling of cover crop growth with thermal units (growing degree days) at the time of this writing.  Analysis of the accessions from the first year of each of the cool season and warm season data have shown that the warm season covers are good candidates for simple GDD regression modeling approaches, and several accessions (‘Tropic Sun’ sunn hemp, Japanese millet, and Chinese red pea, in particular) have high R² values with simple Base 50 GDD models. The cool season covers do not fit into simple GDD models and will require additional model refinement.  This is likely due to significant light limitation during winter months at the Kentucky site where these were evaluated.  GDD models do not account for light limitation, and assume temperature is the only growth limiting factor.  In fact, the daily light integral (DLI) in central Kentucky from late November – late January is growth limiting for many crops.  DLI is further reduced in crops grown in high tunnels, in which each layer of polyethylene plastic reduces light transmission by approximately 10%.    We plan to continue with these activities as we complete the analyses needed to complete the research manuscript for this objective.  Although we do not envision these models to be sufficient to be a stand-alone paper, they provide excellent preliminary data to do region-wide work on evaluating these novel cover crops for high tunnel producers and will be used in future grant proposal activity. 

Novel Cover Crop Outreach.  As a result of this project, and specifically the involvement of our Southern SARE Young Scholar Enhancement award, we were able to connect this work with the Southern Cover Crop Council.  This was an exciting development of the study tour that we took with our YSE scholar, and will provide a home for outreach materials we develop for particular varieties of cover crops.  As we complete our outreach deliverables, we are working on general Extension fact sheets regarding general cover crop considerations for high tunnels, as well as fact sheets on selections for cool season and warm season covers.  We plan to format these for use on the Southern Cover Crops Council website.  Our SARE YSE scholar (Alexandra Tracy) created a draft (Table 1) for warm season cover crop selections based on experimental data and observations from this project, as well as from the literature when supplemental information was needed.