Introduction

Global agriculture is expected to intensify by 2050 to meet the 70% increase in food demand1,2; it must achieve the dual objectives of increased yield on existing farmland3 with reduced environmental degradation4,5. Agricultural intensification produces higher yields from current lands using increased inputs and improved agronomic practices6. One of the most intensified methods of agriculture is raised bed plasticulture (hereafter plasticulture), which involves growing crops on raised plant beds covered with plastic-films7,8. Plasticulture, used with drip irrigation, delivers water and liquid fertilizer precisely to increase water and nutrient use efficiency. The plastic creates a barrier that regulates the soil environment by reducing evaporation, rainfall-driven fertilizer losses, and weeds, controlling temperature, and thereby improving production and crop quality9. With these benefits, global plasticulture usage has increased, particularly in China10, where it grew by 30% (1991–2004) even for lower-value agronomic crops (e.g., corn, wheat)11 in water-scarce regions. Plasticulture has been promoted for its benefits without much understanding of its long-term consequences12.

The rapid expansion of plasticulture has negative environmental impacts owing to increased plastic and fumigant (gaseous toxic pesticide) usage. Plastic films degrade in sunlight, increasing the amount of microplastics in the environment. Microplastics’ prevalence in soil13 and the risk of their uptake into edible plant material through damaged roots14 is an emerging concern. Plastic-covered lands increase runoff15,16 as they reduce infiltration into the soil. Greater nutrient-laden runoff and peak flows adversely affect regional freshwater flows, groundwater recharge17, receiving ecosystems18, and human health19. Fumigants are applied under the plastic to control soilborne biotic crop stressors (plant-parasitic nematodes, bacteria, fungi, weeds), resulting in increased yields20. Fumigants such as chloropicrin (soilborne disease control) and 1,3-dichloropropene (1,3-D; nematode control) are among the top ten globally used pesticides21. Most pesticides are effective against a limited range of pests or restricted due to concerns for human health22 and adverse environmental impacts23. The combined effects of soil heating and pesticide inputs in plasticulture can increase the rate of soil degradation24 and decrease the soil community diversity25. While management of soil pests through pesticides is a necessary component of plasticulture, there are alternative pesticides with greater selectivity and less environmental and human health impacts26. For example, fluensulfone and propamocarb require a “caution” label compared to the “danger, poison” label for a combination of 60% chloropicrin and 40% 1,3-D. However, alternative pesticides are generally cost-prohibitive and therefore not widely used27 in agriculture.

We present a redesigned system to sustainability intensify plasticulture to maintain or increase production28 while reducing the inputs, including the toxic pesticides, and environmental impacts. Redesigned narrower and taller compact bed plasticulture (CBP) synergized with alternative pesticides, was developed to mitigate shortfalls of the conventional plasticulture. Conventional plasticulture practiced by the tomato industry prior to this study, used wide and short beds treated with toxic pesticides. The CBP is a new design developed for this study that could reduce inputs including pesticides, production costs and risks, and environmental impacts and facilitate use of more expensive but less toxic pesticides. We evaluated the environmental and economic effects of CBP and pesticide regimes on biotic (nematode, disease, microbial/fungal community) and abiotic (water, nutrient) stressors and input productivity (water, fertilizer, pesticides, plastic) for tomato, an example high-value and world’s most consumed vegetable crops. Overall, we aimed to determine if CBP and alternative, less toxic pesticide regimes can be a dual strategy that sustainably intensifies plasticulture.

Results and discussion

Redesign for sustainable intensification

Conventional plasticulture beds (CONV; 76 cm wide, 20 cm tall) are less than 50% wetted with a single irrigation drip tape, especially in sandy soils29. Wetted coverage is important for drip-applied water, nutrient, and pesticide (chemigation) efficiencies. To compensate for the lack of bed coverage, long irrigation cycles (>1-hr/cycle, 1.6 mm/hr) are used to increase lateral wetting (Fig. 1). Long irrigation cycles lead to water and nutrients leaching from the root zone (0–30 cm), where 98% of the crop roots are located30 (Fig. 1). The two compact bed designs (Supplementary Table 1), COMP1 (61 cm wide, 25 cm tall) and COMP2 (46 cm wide, 30 cm tall), more closely match the drip-wetted area. Improved wetted coverage during the second season (2017–2018) achieved an 18-day reduction (15 cm) in drought stress period for COMP2 from April 19th to May 12th, 2018, the peak growth and fruit set period for tomato. There was a lack of optimum water at 5 cm from the plants in CONV but not in COMP2 (Fig. 2a), which likely weakened plants between 60 and 80 days after transplanting (DAT) and reduced fruit growth31. Residence time for irrigation water and accompanying nutrients was increased for COMP2 within the root zone, decreasing leaching from this taller bed design. During both seasons, soil moisture (15 cm) in the CONV beds was significantly higher than the COMP2 beds (Fig. 2b). Soil moisture in CONV beds was also above field capacity, a lower threshold of excess water for saturation stress, for more days than COMP2 beds. Beyond plant water stress, excessive moisture also increases water and nutrients losses from the root zone to groundwater and surface water. In addition to increased leaching of drip-applied liquid fertilizer, excessive water also causes increased dissolution and loss of dry fertilizer that is applied at planting.

Fig. 1: Comparison of drip irrigation wetting fronts in conventional and compact plasticulture beds.
Fig. 1: Comparison of drip irrigation wetting fronts in conventional and compact plasticulture beds.
Full size image

Conventional tomato plasticulture uses a wide (76 cm) and short (20 cm tall) bed irrigated with a single drip tape that wets less than half of the bed width. Compact beds (COMP1: 61 cm wide, 25 cm tall; and COMP2: 46 cm wide, 30 cm tall) more closely fit the wetted area. Taller compact beds increase the residence time of water and dissolved nutrients within the plant’s effective root zone (30 cm) and protect the roots from saturation resulting from flooding of the row-middle to result in an increased duration of optimal moisture and retention of nutrients.

Fig. 2: Soil moisture within the CONV (76 cm wide, 20 cm tall) and COMP2 (46 cm wide, 30 cm tall) beds during part of the second season (2017–2018).
Fig. 2: Soil moisture within the CONV (76 cm wide, 20 cm tall) and COMP2 (46 cm wide, 30 cm tall) beds during part of the second season (2017–2018).
Full size image

a COMP2 beds showed higher soil moisture and more plant available water in the top 10 cm during the peak growth and fruit set period compared to CONV. Moisture within CONV beds was closer to the permanent wilting point (0.034 vol/vol), resulting in drought stress. b When moderate rainfall events (Event 1–12 mm rainfall, Event 2–39 mm, and Event 3–20 mm) occurred near fruit set period, the soil moisture 15 cm below bed surface returned more quickly to field capacity (0.134 vol/vol) in the COMP2 beds (~24 h) compared to CONV beds (85–102 h) and reduced the adverse effects of soil saturation on root hypoxia.

A detrimental effect of plasticulture is the increased impervious area of the farm, resulting in higher volume and peak rate of runoff to cause both in-field and off-site flooding. This runoff increases soil erosion and off-site transport of pesticides, dissolved and particulate nitrogen (N) and phosphorus (P), and microplastics that adversely impact the environment (e.g., algae blooms18). Return flow of rainfall from the plastic-mulched (impervious) beds into row middles increases runoff and soil saturation and poses challenges to flood control in the watershed, especially from extreme rainfall. CBP reduces runoff volume by 51–76% compared to CONV32 due to the reduced impervious area and expanded area of non-mulched row-middles, increasing infiltration and water storage, and limiting sediment and nutrient exports. Increased runoff and flooding and soil saturation also reduce oxygen concentrations, damaging roots and decreasing disease resistance, ultimately affecting yields33,34. In-field flooding, a regular occurrence during the growing season, affects tomato yield in as little as 24 h, with permanent damage common in 48 h35. During the second season, COMP2 beds returned from near saturation (0.36 vol/vol) to field capacity (S1: 0.096; S2: 0.13) within 24 h after two moderate rainfall events, while the CONV beds remained above field capacity for more than 72 h (Fig. 2b). In addition, extended periods of flooding and associated saturation in CONV beds occurred between 50 and 60 DAT, close to fruit maturity and harvest, increasing plant stress and susceptibility to soilborne diseases such as the reported Fusarium spp.36. We observed a lower incidence of Fusarium wilt in plants on COMP2 beds (6.7% symptomatic plants) compared to CONV beds (10.7%) during the second season, and this difference was significant (p = 0.04). Thus, abiotic (saturation or drought, Fig. 2) and biotic (Fusarium) stress was reduced in COMP2 beds.

The reduction in bed width decreases the soil volume which could lead to salt burn, a potential tradeoff of CBP. However, the electrical conductivity of the soil near the plant was below the threshold (2.5 ds/m) for salt damage to plant in all treatments in both seasons. Compact beds can reduce nitrate leaching to groundwater as indicated by the nitrate concentration values in soil solution below the root zone (30 cm) were 33–77% lower than CONV in all the four sampling events during the first season (Supplementary Table 3). In one of the sampling events (12/21/2016), the concentration differences were significant (p < 0.03). Additionally, lower soil potassium (K) balance (at planting minus harvest) in COMP2 indicated greater uptake of K compared with CONV within the root zone at 15–30 cm (Season 1) and 0–15 cm depths (Season 2) (Supplementary Table 4). Increased K uptake combined with lower drought and saturation stresses and fusarium wilt contributed to the increased yield in COMP2 for the most desirable, extra-large tomato fruit category (Supplementary Tables 5 and 6), which has a higher market value than smaller sizes. Consistently greater yield for COMP2 in both seasons indicates higher nutrient removal and use efficiency. Overall, CBP provided a more nearly optimal growing environment by collective effects of improved water and nutrient availability and reduced abiotic and biotic stressors.

Synergistic benefits from alternative pesticides

Benefits of CBP can be synergized with more selective pesticides to further improve sustainability. A widely used pesticide for soil disinfestation is the fumigant (FUM; 60% chloropicrin plus 40% 1,3-dichloropropene), which is applied proportional to bed width, i.e., plastic-covered area. Accordingly, narrower COMP2 beds received 40% less pesticide than CONV. With CBP, less hazardous but more expensive, alternative pesticide regimes (ALT) of chloropicrin paired with fluensulfone became economically analogous. For example, the cost (US, 2018 dollar) of pesticide (Table 1) for the ALT regime for COMP1 was $1352/ ha, similar to the CONV (FUM regime, $1354/ha). Additionally, the increased drip-wetted coverage in CBP compared to CONV allows for the application of drip-applied non-fumigant (NOFUM) pesticides (fluensulfone and propamocarb) which have difficulty covering more expansive CONV beds. Applying less toxic pesticides (NOFUM), through the drip tape limits human contact and potential injury to farmworkers and grower’s liability during application since they are applied remotely at water and fertilizer pump station. FUM, by contrast, was injected through chisels mounted on a tractor-driven bedder at bedding. NOFUM enables growers more flexibility because they can plant within seven days after application compared to FUM, which has a 21-day re-entry period. NOFUM is also more selective than FUM, allowing populations of beneficial soil microbes and nematodes to survive37.

Table 1 Partial budget analysis to determine the economic impact of conversion from the CONV (76 cm wide, 20 cm tall) beds treated with the conventional FUM (60% chloropicrin + 40% 1,3-dichloroproene) to the alternative treatments
Full size table

Increased microbial diversity affects soil nutrient cycling and health38. The dual strategy of compact geometry (COMP2) and pesticide regime (ALT) positively impacted both soil bacterial and fungal community diversity and composition. The bacterial diversity was significantly greater in COMP beds treated with ALT and FUM, while it had significantly lower fungal diversity compared to NOFUM treatments (Fig. 3). Further, in the COMP beds treated with ALT and FUM, four genera belonging to the phylum Firmicutes were significantly enriched at the 0–15 cm depth. Some of the bacteria in these genera have been found to have plant growth promoting properties39. Additionally, three fungal genera belonging to the phylum Ascomycota, containing Fusarium spp., were depleted in soils treated with ALT and FUM in both CONV and COMP treatments (Fig. 3). This depletion of harmful fungal phylum, combined with the reduced drought and saturation stress in COMP2, reduced the risk of disease incidence observed for both COMP2 (ALT and FUM) treatments.

Fig. 3: Heat map of relative abundance of bacterial and fungal genera during the second season (2017–2018).
Fig. 3: Heat map of relative abundance of bacterial and fungal genera during the second season (2017–2018).
Full size image

Heat maps showing changes in the relative abundance of the bacterial and fungal genera found to be significantly enriched or depleted (p < 0.05) in soil samples taken at two soil depths (0–15 cm and 15–30 cm) near transplant and at the end of harvest from conventional (CONV; 76 cm wide, 20 cm tall) and compact (COMP2; 46 cm wide, 30 cm tall) beds treated with pesticide formulations including fumigants FUM (60% chloropicrin + 40% 1,3-dicholorpropene) and ALT (chloropicrin + fluensulfone) in comparison to NOFUM (fluensulfone+ propamocarb) treatment for the second season.

The decreased incidence of Fusarium wilt observed in COMP2 beds was influenced by the increase in the population of root-knot nematode (RKN, Meloidogyne spp.). The pathological complex between the presence of RKN and reduced resistance of plants to Fusarium spp. is well-established40. We observed increased FUM application efficiency with reduced bed width during the first season. By the end-of-harvest, root gall ratings, a cumulative measure of RKN damage, were significantly reduced for two bed sizes (COMP1 and COMP2) for FUM (Fig. 4a). The population of RKN decreased with bed width (Fig. 4b) despite receiving up to 40% less FUM, a combined effect of increased height of the fumigated bed and better mixing and coverage of FUM due to reduced bed width. Taller COMP beds required nematodes to travel farther from untreated soils below beds to feed on plant roots, as evidenced by the lower root galling index (Fig. 4a). The use of alternative pesticide regimes alone is an effective strategy for improving soil health and reducing the risk of soilborne pests. Together, COMP2 combined with alternative pesticides resulted in synergistic benefits of reduced pesticide usage while still reducing risks from RKN and Fusarium wilt pathogens.

Fig. 4: Bar graphs showing indicators of nematode activity for the first season (2016–2017).
Fig. 4: Bar graphs showing indicators of nematode activity for the first season (2016–2017).
Full size image

a Mean root gall index and b root knot nematode (RKN) population at the end of harvest for the first season. Error bars indicate standard deviations. For FUM (60% chloropicrin and 40% 1,3-dichloropropene) treated CONV (76 cm wide, 20 cm tall; Fig. 1) and COMP (COMP1: 61 cm wide, 25 cm tall; COMP2: 46 cm wide, 30 cm tall; Fig. 1) beds, root galling index and nematode count were significantly lower (p = 0.01) for COMP1 and COMP2 beds despite using less pre-plant pesticides than CONV. This significant difference is not seen in the ALT (100% chloropicrin and fluensulfone) treatments indicating that the application method (ALT: drip-applied and FUM: chisel-applied) can increase soil coverage.

Outcomes from dual sustainable intensification strategy

The dual strategy of an optimal growing environment with a compact bed (COMP2) and alternative pesticide regimes (ALT) increased large, high-valued, late-season fruit production during both seasons (Supplementary Tables 5 and 6). Increased late-season yield resulted in higher net benefits (revenue minus costs that varied among treatments) for both seasons (Table 1). For the first season, higher yield for COMP2-FUM and COMP2-ALT combined with reduced plastic and pesticide costs increased net profit by $1266 and $1134/ha, respectively. For second season, COMP2-ALT provided greater net profit ($4462/ha) due to similar increases in yields. For the ALT regime, greater net profit was also observed for the CONV-ALT ($3708/ha). COMP2-ALT produced a greater amount of large, high-value fruit during both seasons, which increased the net benefit by an average of $2798/ha. The increased extra-large, late-season fruit for COMP2-ALT reduces the late-season risk in case of increased sale prices without significant additional inputs beyond crop maintenance (irrigation, fertigation, and herbicide/fungicide).

A consistent benefit of the dual strategy of CBP and ALT is the reduced input costs due to reduced bed width. The use of COMP2-FUM, COMP2-ALT, and COMP2-NOFUM treatments reduced input costs by $536, $403, and $683 per ha, respectively. This reduction occurs regardless of the market price, which is highly variable and dependent on external forces. This consistent decrease in input cost can easily offset the cost of new/modified machinery to make compact beds, a one-time investment of US$2000–20,000. This investment would be paid off in one year for farms larger than 50 ha, typical landholding of farms in North America41. Adopting the dual strategy helps growers reduce production costs and risks to produce more with less.

Individually or combined, CBP (COMP2) and ALT achieve sustainable intensification to increase the system productivity (yield per unit of fertilizer, water, or plastic). For both seasons, COMP2 treatments (COMP2-FUM and COMP2-ALT) increased N, P, and water productivity by almost 4% compared with the conventional treatment (CONV-FUM) (Fig. 5). COMP2-ALT and CONV-ALT treatments increased N, P, and water productivity by almost 12%. The increased input productivity is especially important for N and P, as they are limiting nutrients for freshwater systems globally42. Increased P productivity also helps alleviate predicted global shortage of P fertilizer43.

Fig. 5: Graphs showing the ranks of conventional and compact systems in six sustainability metrics averaged over both seasons.
Fig. 5: Graphs showing the ranks of conventional and compact systems in six sustainability metrics averaged over both seasons.
Full size image

Comparison of the eight treatments (CONV and COMP) based on six sustainability metrics (water [mm/kg of tomato], nitrogen [kg N/kg tomato], phosphorus [kg P/kg tomato], plastic [kg plastic/ kg tomato], fumigant [% reduction/ kg tomato], and cost [$/kg tomato]) averaged over the two seasons. Larger shaded area indicates higher productivity of the system. The coloration of figures shows the gradient of desirability (red less desirable, green more desirable) for each treatment. The CONV (76 cm wide, 20 cm tall) beds were ranked lower for most categories particularly the plastic and fumigant. The COMP (COMP1: 61 cm wide, 25 cm tall; COMP2: 46 cm wide, 30 cm tall) beds were ranked highest in the water [mm/kg of tomato], nitrogen [kg N/kg tomato], phosphorus [kg P/ kg tomato], plastic [kg plastic/ kg tomato], fumigant [% reduction/kg tomato], and cost [$/kg tomato] categories. The COMP2 beds treated with FUM (60% chloropicrin and 40% 1,3-dichloropropene) and ALT (100% chloropicrin and fluensulfone) were ranked highest for productivity. All bed sizes treated with the NOFUM (fluensulfone and propamocarb) pesticide regime were ranked lowest.

Increased productivity of water, fertilizer (N, P), pesticide, and plastic results in a sustainably intensified production system (Fig. 5) by reducing these inputs to produce the same or higher amount of high-value fresh fruits/vegetables. COMP2 treatments (COMP2- FUM, COMP2-ALT) increased the plastic productivity by almost 15% in the first season, producing more with less plastic. During the second season, COMP2-ALT and COMP2-FUM increased plastic productivity by almost 24% and 11%, respectively, compared with the CONV-FUM treatments. The amount of agricultural plastics and associated waste is a serious concern12, particularly microplastics in the terrestrial environment. Plastics release greenhouse gases (GHG) into the atmosphere during production, application, removal, and disposal to landfills (or recycling) which directly contribute to global warming. Field burning of plastic is common, but when restricted by law, the plastic is either transported to a landfill ($99/ha, Table 1) or a recycling facility, resulting in additional harmful emissions and a cost to the grower. This cost becomes more important in regions with stricter environmental laws (e.g., California, USA). CBP alone can reduce the carbon footprint by 5% for COMP1-FUM and 10% for COMP2-FUM compared with the CONV-FUM treatment44. Further GHG emission reductions are achieved with CBP and ALT due to lower plastic and energy usage, resulting in lessening the plasticulture’s contribution to GHG emissions and plastic pollution.

Overall, the dual strategy of redesigned soil growing environment (CBP) and less toxic pesticide (ALT) maintained or increased yields while reducing inputs and negative environmental impacts. The COMP2-ALT and COMP2-FUM treatments were among the top three ranked treatments (8 = highest/best, 1 = lowest/worst) for each of the six productivity metrics (N, P, % reduction in pesticide, plastic, and water productivity, and cost reduction) shown in the sustainability graphs (Fig. 5). The NOFUM treatments (CONV-NOFUM and COMP2-NOFUM) were the lowest-ranked in all categories besides costs (Fig. 5). The CONV treatments (CONV-FUM, CONV-ALT, CONV-NOFUM) were ranked lower than CBP treatments due to increased inputs of plastics and pesticides necessary for the wider CONV beds. Additional environmental benefits specific to COMP2 treatments include reduced losses of water, nutrients, and pesticides to regional ecosystems from runoff, leaching to groundwater, and carbon emissions44.

Economic and environmental benefits from this study have already benefited North American growers. As result of this study, two of the ten largest tomato growers in North America, one of whom participated in this study, have already adopted CBP (COMP1) to reduce inputs and costs, increase system efficiency, and reduce their environmental footprint across the US and Mexico. There have been further reports of growers in the US adopting CBP for other crops such as pepper and watermelon that use high inputs to increase production and meet cosmetic market standards. Growers are further exploring conversion to COMP2 beds due to 60% or more increase in energy, fertilizer, and pesticides costs for 2022 compared 202145. Conversion to COMP2 treatments on 10,000 ha would increase system productivity, especially for energy-intensive inputs, by reducing plastic usage by 220 tonnes and pesticide usage by 2438 tonnes while increasing annual income by nearly $28 million. Approximately 200,000 ha of tomato is planted in North America alone each year46. Complete conversion to CBP would result in a 4.4 Mt reduction in plastic usage, a 22.8 Mt reduction in pre-plant pesticide usage, and a $183 million increase in annual income. When scaled up for other plasticulture crops in the US, including cucurbits, peppers, and strawberries on 312,126 ha47: the benefits increase to 6.9 Mt less plastic, 35.5 Mt less pre-plant pesticides, and $385 million increased income.

The redesign of plasticulture to CBP can also help minimize some negative impacts of intensive agriculture, moving closer to a sustainable system4. Reducing inputs with the redesign and adoption of the dual strategy for sustainable intensification of tomato production will reduce two of the top ten most widely used pesticides globally21 and have a positive impact on production and the ecosystem48. The cost reduction from CBP may sufficiently incentivize growers to use more costly biodegradable mulches. Fumigants are not currently permitted to be used with biodegradable plastic due to its high permeability; however non-fumigant pesticides are. CBP allows for transitioning to a system that combines non-fumigant pesticides and biodegradable mulch with an added benefit of enhanced soil biodiversity. The redesigned beds allow the plasticulture industry to move towards a carbon-neutral system. Global adoption of CBP will help meet the GHG emission goals necessary to limit climate change49. Reduction in carbon footprint may become particularly valuable to growers once ongoing legislative efforts on establishing a carbon credit market in Europe and US become a reality.

CBP increases global food security by mitigating environmental impacts and acting as a climate change adaptation strategy, reducing risks from weather extremes (drought and flooding)49. CBP increased root zone soil moisture during dry periods, reducing irrigation withdrawals in water-scarce regions (e.g., Southeast and Southwest US). The increase in global food demand and water scarcity from changing climate50 is likely to expand plasticulture globally. For temperate and tropical regions with moderate rainfall, reduced impervious area with CBP will increase in-field storage of rainfall and groundwater recharge while reducing runoff, flooding, and associated soil loss. Irrespective of reduced flooding, taller CBP protect crops by keeping plant roots higher in the bed, reducing saturation stress, associated diseases, and nutrient (e.g., N and P) losses to surface and ground waters. Reduced flooding risk is particularly important as climate change is predicted to increase rainfall intensity51 and pressure on regional water management systems17. Reduced energy use and costs to pump irrigation and drainage with CBP make it an economic climate change mitigation measure. The combined reduction by CBP in environmental impacts, risks from pests/diseases, and extreme weather will increase food security and increase the system’s resiliency.

Climate change will likely alter the endemic ranges and disease severity52 of common plant pathogens. Projected increases in seasonal rainfall variability and rainfall intensity are likely to increase the risk of water-vectored plant diseases, like Fusarium spp. The reduced in-bed soil moisture and decreased rootzone saturation period with the reduction in Fusarium spp., observed in COMP2 treatments, will act as an additional climate change adaptation strategy for plasticulture. Additionally, higher temperatures seen with climate change will increase nematode reproductive potential and affect the viability of plant varieties genetically resistant to RKN, increasing their risk of infection53. These disease benefits might become more critical as the effects of climate change become increasingly evident, exacerbating the effects of both Fusarium spp. and Meloidogyne spp., two crop stressors with major impacts on the global food production.

Future demand for fruit and vegetables will increase, resulting in additional land conversion, which is detrimental to the surrounding landscape and adjacent ecosystems54. CBP allows further system intensification by decreasing row-to-row spacing and using the extra space available to increase plant population (33% for COMP2) without additional land clearing/conversion. Reduced need for agricultural land limits the competition for land associated with deforestation for agriculture and housing which is particularly important in ecologically sensitive, coastal regions such as the Everglades and the Chesapeake Bay watersheds in the US. As the urban areas expand to farmlands, pesticide use is being limited due to health concerns, particularly in peri-urban areas. The use of alternative pesticides can meet these regulations cost-effectively with CBP. Together, CBP with alternative pesticides can sustainably intensify plasticulture where costs are high, restrictions on pesticides exist, or land use conversion has negative economic or environmental consequences.

Evaluation of multiple facets of the plasticulture production system was necessary to determine the complex and interconnected nature of the redesigned CBP systems and evaluate the conventional system’s environmental impacts. Future studies should field-verify these results for other crops (e.g., cucurbits) and soil-climatic regions of the world. Future efforts should also be made to develop bed geometry specific irrigation management to evaluate its water conservation potential. Redesigned plasticulture production system will help improve long-term food security, land use efficiency, and sustainability in a more interconnected economy confronted with the looming challenge of climate change. Our results show that a redesign of the current plasticulture to CBP, developed with soil-water interactions at the forefront, synergized with alternative, more selective pesticide (ALT), increases productivity with sustainable intensification.

Methods

Experimental areas, bed designs, and fumigation regimes

Sites and cropping systems

Two experiments on round tomato (Solanum lycopersicum, variety: Harris Moran 8849) were conducted on commercial farms, the first occurred during the fall growing season of 2016–2017, and the second occurred during the spring season of 2017–2018. The first season was conducted on a farm in the ecologically sensitive Everglades basin, Florida, that is faced with challenges of flooding and nutrient loads, especially phosphorus. The second experiment was conducted on a farm located in south-central Florida in the Manatee River basin that is impacted by both phosphorus and nitrogen discharges and contains two run-of-river water supply reservoirs. The soil at the first experimental site is classified as Basinger fine sand and the soil at the second site is classified as Waveland fine sand55.

The fall season (Sep–Jan) represents a gradually cooling season with a mean monthly temperature of 24.9 °C at planting and 17.2 °C at the time of harvest. The spring (Jan–Jun.) is a gradually warming season with a mean monthly temperature of 16.1 °C at planting and 27.6 °C at harvest. The average annual rainfall for the region during the fall is 373 mm and during the spring it is 541 mm. Observed rainfall during the experiment was 93 mm and 536 mm for the fall and spring seasons, respectively. The evaporative demand reflected the climatic conditions, as the spring season had higher potential ET (467 mm) than the winter season (312 mm). Together, these two growing seasons represent typical growing environments in temperate, sub-tropical, and semi-arid regions of North America and other continents56.

Tomato seedlings were transplanted at the center of the bed in a single line, six weeks after covering the pesticide treated beds with plastic, a common production practice for plasticulture. Plants were spaced 61 cm apart for the first season, and for the second season, plant spacing was reduced to 51 cm. This alteration in plant spacing is common in commercial tomato production areas of North America and other regions due to differences in seasonal/environmental conditions and land availability. All cultural practices, including harvesting, post-planting pest control, irrigation timing and duration, were same across all bed geometries and pre-plant pesticide regimes, and were carried out by the grower-cooperator.

Bed geometries and pre-plant pesticide regimes

During the first season, three-bed geometries, one conventional (CONV: 76 cm wide, 20 cm tall) and two compact (COMP1: 61 cm wide, 25 cm tall; COMP2: 46 cm wide, 30 cm tall), were evaluated (Supplementary Table 1). Two bed geometries (CONV and COMP2) were evaluated for the second season. The formation of plastic mulched beds is a four-stage operation comprised of fertilizer and fumigant applications followed by drip tape and plastic mulch placement44. The first stage involved land preparation, including soil disking and in-bed fertilizer application at a rate (kg/ha) of 18N-54P-34K. For the second stage, a bedding machine with the appropriate press pan for the selected bed geometry formed the beds, injected fumigant with narrow knife-like shanks, and applied fertilizer (average 212N-0P-345K kg/ha) in two bands at the top of the bed. The total amount of fertilizer applied was the same for all the treatments. The third stage involved the application of pre-emergent herbicide [S-metolachlor], which was applied to the pre-formed beds. Lastly, a plastic laying machine fitted with a drip-tape roll carrier laid the drip tape (0.91 L h-1/emitter, 30 cm emitter spacing) before laying the metalized plastic mulch (0.028 mm), which covered the tops and sides of the beds.

During the first season, two pre-plant pesticide regimes were evaluated: the conventional fumigation (FUM) of 314 kg of Pic60 (60% chloropicrin and 40% 1,3-D) per broadcast-ha, and an alternative (ALT) of 177 kg of 99% chloropicrin plus 4.15 L of fluensulfone per broadcast-ha (Supplementary Table 1). For the second season, a non-fumigant (NOFUM) regime with 4.15 L of fluensulfone and 2.23 L of propamocarb per broadcast-ha was added to determine if pest control was possible without using fumigants (Supplementary Table 1). The amount of pre-plant pesticides applied to each treatment was reduced based on the bottom width of the bed resulting in variable amount of pre-plant pesticides (Supplementary Table 1). Fluensulfone (S1 and S2) and propamocarb (S2) were applied through drip system, seven days before transplanting the tomato seedlings. Pesticide (NOFUM) application seven days before transplant allowed the grower additional flexibility in planting time and increased the length of pest control compared to the conventional fumigant pesticides which restrict planting to at least three weeks after fumigation. NOFUM pesticides (fluensulfone and propamocarb) were applied using an external water tank and injection pump, which supplied water and pesticides directly to drip tape lines. Due to the phytotoxic properties of the fluensulfone, 7.6 mm of irrigation was applied to flush any residual out of the root zone over a two-day period after injection. As per the recommended (label) application rate, the volume of all pre-plant pesticides (fumigant and non-fumigant) was based on the width of the plastic-covered bed. Therefore, COMP1 and COMP2 received 20% and 40% less pesticide, than the CONV beds (Supplementary Table 1).

Experimental design and field layout

During the first season, a randomized block design was used. The field was blocked north to south to represent dual-sided slope from the center to the ends of the field. Irrigation risers were located at the center of the field (Supplementary Fig. 1A). Each experimental unit (treatment-replication) consisted of a three-bed grouping, each with a length of 100 m and 0.18 ha area. There were four replications for each of the six treatments. For the second season, an incomplete randomized strip plot design was implemented (Supplementary Fig. 1B). The field was blocked (north and south), similar to the first season. Four replications were implemented for each of the six treatments, with each replication consisting of a single bed with a length of 70 m and an area of 0.01 ha.

Data collection

Timelines of data collection and cultural practices are presented in Supplementary Table 2.

Soil moisture and electrical conductivity

Soil moisture sensors, CS625 [Campbell Scientific, Logan, UT, US], were installed vertically (0–10 cm) and horizontally into the side of the bed, 15 cm and 30 cm below the bed surface (Fig. 2), to measure water content every 15-minutes. Root zone soil moisture and electrical conductivity were monitored at 15-min frequency by placing a sensor (Model CS655, Campbell Scientific, Logan, UT, US) 5 cm from the plant’s root collar, with the plant located between the sensor and the drip tape. Electrical conductivity values were used as a surrogate measure of salinity (dissolved salts) within the root zone to determine if salt concentrations reached damaging levels in the compact beds due to reduced soil volume in COMP1 and COMP2 beds.

Irrigation monitoring

Irrigation timing, duration and volume was monitored every 15 min with an in-line impeller flowmeter connected to a datalogger. Irrigation management was consistent between all treatments after planting. A single drip tape (emitter spacing 30 cm) with a flow rate of 0.91 liters per hour per emitter was used for both seasons. During the first season, irrigation continued until the last harvest. However, during the latter portion of the second growing season, multiple rainfall events resulted in irrigation being terminated two days after the first harvest. Flooding frequency and duration were environmental risks that were evaluated in this study. Extent of flooding and saturation stress was determined by the length of time when the measured soil moisture was above the field capacity (soil moisture at −0.33 bar).

Soil solution nutrient sampling

Soil solution (pore water) samples were collected 3 to 4 times during each growing season using 5-cm diameter suction lysimeters with ceramic tips [Soil Moisture Equipment Corp., Santa Barbara, CA, US] installed 15 and 30 cm below the bed surface in each of the FUM and ALT treatments. (NO3-N + NH4-N) Soil solution samples were analyzed by the Analytical Research Laboratory (ARL), University of Florida (UF), Institute of Food and Agricultural Sciences (IFAS), Gainesville, FL, USA for dissolved inorganic nitrogen (NO3-N + NH4-N) species, nitrate (NO3-N; EPA 353.2) and ammonium (NH4-N; EPA 350.1). The 15 cm samples were used to evaluate if nutrient concentrations within root zone differed between treatments. The 30 cm samples were used to evaluate if there were differences in nutrients leaching below the root zone for each treatment.

Soil Matrix

Soil samples collected at planting and after the final harvest of the season (Supplementary Table 2) were analyzed for nutrients (N,P,K), nematode abundance, and microbial community composition. Samples were taken using a split core soil sampler (length = 30 cm., dia. = 5 cm) that housed an acetate liner. Each sample analyzed was an aggregate of the eight-core samples, with four samples collected along the plant line and four across the width of the bed. Core samples were divided into two depths, 0–15 cm and 15–30 cm. These samples were air-dried and sent to UF/IFAS ARL for nutrient content analyses. At the lab, samples were digested, and the resulting solution was analyzed for concentrations of total P (TP; EPA 365.1), K (EPA 200.7), NH4-N (NH4-N; EPA 350.1), NO3-N (NO3-N EPA 353.2), and total Kjeldahl nitrogen (TKN; EPA 357.2).

Yield

The yield was measured from ten consecutive representative plants in each treatment-replication. Tomato was harvested multiple times, thrice for first and twice for the second season (Supplementary Table 2). Tomatoes were graded into three categories as per the fruit size standards set by the United States Department of Agriculture (USDA): extra-large (>7.00 cm), large (6.35–7.00 cm), and medium (5.72–6.35 cm)57. Total marketable yield excluded the culled category (scarred, sunburnt, or blemished), which are not considered marketable. Yields from multiple harvests from each treatment replication were aggregated to calculate the total marketable yield.

Nematode

Soil in the rhizosphere, the soil region near plant roots, was collected during each season from a depth of 30 cm after final harvest and analyzed for nematodes. For the second season, additional soil samples were collected at the time of planting from 15 cm as well as 30 cm depths. Soil samples (200 cc) were extracted using a modified Baermann method58 (salad spinner method). Both plant-parasitic (Root-knot nematode RKN; Meloidogyne spp.) and non-plant-parasitic nematodes (bacterivores, fungivores, and omnivores) were quantified using an inverted microscope. Root gall assessment was conducted on eight uprooted plants in each treatment replication using a 0–10 scale59 where 0 indicates no visible root galling (no damage from RKN) and 10 represents 100% galling with no visible fibrous roots, indicating maximum threat to plants from parasitic nematodes.

Soil community composition

Total DNA from 0.25 g of soil samples collected during the second season was extracted using the DNeasy PowerSoil PowerLyzer Kit [Qiagen, Germantown, MD, US]. The extracted DNA was quantified using a Qubit Fluorometer (Thermofisher Scientific) through the Quant-iT dsDNA HS Assay Kit and was sent for sequencing to the DNA Services Facility at the University of Illinois, Chicago, IL, US. The V4 region of the bacterial 16S rRNA gene was amplified using 515Fa and 926 R primers60, and the fungal ITS region using ITS1F/ITS2R primers61.

Raw sequence reads were analyzed using QIIME2 v2018.462,63. Reads were dereplicated using DADA264, resulting in amplicon sequence variant (ASV) tables containing read counts. The representative bacterial and fungal ASVs were assigned to the SILVA 12865 and UNITE 7.2 databases66, respectively, using naïve Bayes classifier in QIIME 267. Raw reads used in this study can be found in the National Center for Biotechnology Information Sequence Read Archive under accession number PRJNA672958.

Alpha and beta diversity analyses were performed using the R package Phyloseq on log-normalized data68. Bacterial alpha diversity was estimated by calculating bacterial richness (number of observed ASVs), Shannon, and inverse Simpson indices. Bacterial and fungal beta diversity were estimated according to a principal coordinate analysis (PCoA) on Bray-Curtis distance between samples. Subsequently, a four-way Permutational Analysis of Variance (PERMANOVA)69 was performed to determine significant differences in bacterial beta diversity between samples based on the bed size (CONV and COMP2), pesticide regime (NOFUM, ALT, and FUM), time point (at planting and harvest), and soil depth (0–15, 15–30 cm) using the ‘adonis()’ function of the R package ‘vegan’ with 999 permutations. A two-way PERMANOVA was used to investigate the effect of bed size (CONV and COMP2), treatment (NOFUM, ALT, and FUM), and their interaction on the bacterial and fungal community structure at the phylum level using the ‘vegan’ package.

Based on the relative abundance profiles of bacterial and fungal ASVs, genera with significantly differential relative abundances between the ALT and FUM treated soils vs. NOFUM were determined using paired DESeq2 comparison analysis70. The read count matrix for DESeq2 testing was normalized using the DESeqVS method described by Weiss et al. (2017). P-values were corrected using the Benjamini–Hochberg method in DESeq2, and values lower than 0.05 were considered significant. Heat maps (Fig. 3) were generated, including the relative abundances of the significantly enriched or depleted taxa using the ‘pheatmap’ package in R software.

Plant disease incidence

Observations from the first season indicated that difference in the occurrences of plant disease among treatments, especially Fusarium wilt (Fusarium spp.) may occur given the differences in soil saturation between bed geometries. Consequently, disease incidence (percentage of plants exhibiting disease symptoms of Fusarium spp.) was measured for the second season 4 days before the final harvest by surveying all 136 plants in each treatment replication.

Systems analysis

Statistical analyses

Yield, nutrient concentrations (soil matrix and solution), nematode count, root gall index, and disease severity were statistically analyzed to detect treatment effect. Variables in each category were grouped according to date and depth, when applicable (e.g., soil solution), and each group from each season was checked for normality, homogeneity of variance, and independence. Then each group was analyzed separately using two-way ANOVA with factors of bed geometry, fumigant type, and their interactions. The differences between treatments were checked using the Tukey-Kramer honest significant difference test at the significance levels of α = 0.05.

Partial economic budgeting

An enterprise budget71 for the fresh market production of tomato was used to create a sub-list of costs that varied among treatments. Any differences between treatments and CONV-FUM were recorded during the growing seasons. The costs that vary among treatments included pesticide regime costs, plastic mulch, dumpster contract (plastic mulch disposal to landfill), and yield (when statistically different). Pesticide costs were based on estimates collected from sales quotes, and the amount of pesticide applied to the beds, proportional to the bed’s width. The cost of plastic mulch and field clean-up were estimated to be proportional to the surface area of plastic used for each bed geometry. The expected change in revenue was calculated by averaging increased yield categories and multiplying them by the field price. The gross revenue was calculated using the price at the nearest USDA shipping point for Central and South Florida (Orlando, USA) for the day of harvest. The harvesting costs included labor, packing, and shipping for each 11.4 kg unit (25 lb box). The net benefit was then generated by subtracting the costs that varied and the harvesting costs from the gross revenue. The net benefit from each treatment was then compared to the industry standard CONV-FUM to determine the economic benefit of adopting compact beds with FUM or ALT treatments.

System productivity

Productivity is an important metric and can be used to determine if increased intensification benefits the overall production system. Six sustainability metrics (water, N and P fertilizers, plastic, pesticide, and cost savings) were used to evaluate eight production systems (Fig. 5). Four system productivity metrics were calculated by dividing the output, yield (kg of tomatoes) by kg of applied N, P, and plastic72. The productivity of water was calculated per liter of applied irrigation. The metric of reduction in pesticide was dependent on the width of the bed bottom regardless of the type of pre-plant pesticide applied. Accordingly, % reduction in pesticide was used to compare the pesticide alternatives. Greater the percent reduction, the less pesticides were used in the system. The individual input productivities and cost savings were ranked for each of the eight treatments (CONV-FUM, CONV-ALT, CONV-NOFUM, COMP1-FUM, COMP1-ALT, COMP2-FUM, COMP2-ALT, and COMP2-NOFUM) with the productivities averaged across the two seasons where available. The six metrics were plotted in Fig. 5 to provide a holistic assessment through visual comparison of overall sustainability of the eight treatments where the larger the polygon size, the higher is the productivity.