ORIGINAL RESEARCH published: 12 November 2020 doi: 10.3389/fpls.2020.596550 Closed-Loop Crop Cascade to Optimize Nutrient Flows and Grow Low-Impact Vegetables in Cities Martí Rufí-Salís 1,2* , Felipe Parada 1 , Verónica Arcas-Pilz 1 , Anna Petit-Boix 3 , Gara Villalba 1,2 and Xavier Gabarrell 1,2 1 Sostenipra Research Group (2017 SGR 1683), María de Maeztu Unit, CEX2019-000940-M, Institut de Ciència i Tecnologia Ambientals (ICTA-UAB), Universitat Autònoma de Barcelona, Barcelona, Spain, 2 Department of Chemical, Biological, and Environmental Engineering, Universitat Autònoma de Barcelona, Barcelona, Spain, 3 Chair of Societal Transition and Circular Economy, University of Freiburg, Freiburg, Germany Urban agriculture systems can significantly contribute towards mitigating the impacts of inefficient and complex food supply chains and increase urban food sovereignty. Moreover, improving these urban agriculture systems in terms of nutrient management can lead to a better environmental performance. Based on a rooftop greenhouse in the Barcelona region, we propose a cascade system where the leachates of a tomato cycle from January to July (donor crop) are used as the main irrigation source for five successive lettuce cycles (receiving crop). By determining the agronomic performance Edited by: and the nutrient metabolism of the system, we aimed to define the potential of these Md Asaduzzaman, systems to avoid nutrient depletion and mitigate eutrophication, while scaling the system Bangladesh Agricultural Research Institute, Bangladesh in terms of nutrient supply between the donor and the receiving crops. The results Reviewed by: showed that low yields (below 130 g per lettuce plant) are obtained if a cascade system Niels Holst, is used during the early stage of the donor crop, as the amount of nutrients in donor’s Aarhus University, Denmark Leopoldo Mendoza Espinosa, leachates, specially N (62.4 mg irrigated per plant in the first cycle), was not enough to Universidad Autónoma de Baja feed the lettuce receiving crop. This effect was also observed in the nutrient content of California, Mexico the lettuce, which increased with every test until equaling the control (4.4% of N content) *Correspondence: as the leachates got richer, although too high electrical conductivity values (near 3 dS/m) Martí Rufí-Salís Marti.Rufi@uab.cat were reached at the end of the donor crop cycle. Findings on the uptake of the residual nutrient flows showed how the cascade system was able to take advantage of the Specialty section: nutrients to produce local lettuce while mitigating the effect of N and P in the freshwater This article was submitted to Crop and Product Physiology, and marine environments. Considering our case study, we finally quantified the scale a section of the journal between the donor and receiving crops and proposed three major ideas to optimize the Frontiers in Plant Science nutrient flows while maintaining the yield and quality of the vegetables produced in the Received: 19 August 2020 Accepted: 26 October 2020 receiving crop. Published: 12 November 2020 Keywords: cascade systems, nutrient recycling, urban agriculture, industrial ecology, urban metabolism Citation: Rufí-Salís M, Parada F, Arcas-Pilz V, Petit-Boix A, Villalba G and Gabarrell X (2020) Closed-Loop INTRODUCTION Crop Cascade to Optimize Nutrient Flows and Grow Low-Impact Cities cover 3% of the Earth’s surface area, but host 55% of the world’s population (United Nations, Vegetables in Cities. 2014; SEDAC, 2016). The global nutritional demand thus concentrates in urban areas, resulting Front. Plant Sci. 11:596550. in long and complex supply chains. Urban agriculture (UA) has arisen as a promising solution doi: 10.3389/fpls.2020.596550 that brings production closer to consumption points. However, UA is not free of environmental Frontiers in Plant Science | www.frontiersin.org 1 November 2020 | Volume 11 | Article 596550 Rufí-Salís et al. Cascade System in Urban Agriculture impacts. Similar to conventional agricultural systems, nutrient crops) to produce vegetables while mitigating nutrient depletion discharge resulting from intensive fertilizer use is particularly through the use of a cascade system and the quantification of problematic due its impact on water eutrophication (Muñoz the nutrient flows. et al., 2008; Torrellas et al., 2012; Romero-Gámez et al., 2014, The present article aims to tap the full potential of cascade 2012; Boneta et al., 2019). The application of circular economy systems to produce local-grown vegetables in the framework of strategies in UA systems could help mitigate these impacts. UA while diminishing the nutrient load by closing nutrient cycles Closing the nutrient cycles in soilless systems maintains the utility in the framework of urban metabolism. Based on the agronomic and value of scarce resources (Bocken et al., 2017), produces and nutritional performance of five successive receiving crops a regenerative effect on the environment (Ellen MacArthur of lettuce (Lactuca sativa, Maravilla) irrigated by a long-cycle Foundation and McKinsey Center for Business and Environment, donor crop of tomato (Lycopersicon esculentum, Arawak), we 2015), and contributes to a reduction in water and fertilizer determined the feasibility and nutritional implications of cascade consumption (Carmassi et al., 2005; Rufí-Salís et al., 2020). systems and provide recommendations to further improve Two main strategies can be found in the literature to close the performance of this setup. This will enable a better the nutrient cycles in soilless systems. First, recirculation systems implementation of UA systems in cities and inform decision- use drained water and nutrients in the same crop. This setup makers about the main benefits and improvement potential helps to reduce water and nutrient losses by closing resource of reusing nutrients in UA by considering the principles of a flows (Ahmed et al., 2000; Agung Putra and Yuliando, 2015). circular economy. However, it demands additional infrastructure, which requires high investment costs (De Pascale and Maggio, 2005; García- Caparrós et al., 2018) and increases the environmental impacts MATERIALS AND METHODS of the system (Rufí-Salís et al., 2020). Moreover, the high salinity of the drained water can cause yield reduction if it is directly System Under Study, Crop Description, recirculated to the crop without treatment or partial discharge and Experimental Design (Magán et al., 2008; Grewal et al., 2011). The study was performed in a rooftop greenhouse (RTG) located Second, cascade systems use drained water and nutrients to on the ICTA-ICP building (41.497681N, 2.108834E) on the irrigate another crop (Incrocci et al., 2003). Since nutrients tend campus of the Universitat Autònoma de Barcelona, 15 km west to reach high concentrations in the leachates and thus increase of Barcelona, in the West Mediterranean region of the Iberian the salinity of this water flow, the salt tolerance of the receiving Peninsula. The crops grown in the greenhouse used rainwater crop has been the main field of inquiry in the literature. For harvested from a roof surface of 400 m2 (plus 500 m2 from the instance, Shannon et al. (2000) analyzed the salt tolerance of neighboring building). The irrigation system was hydroponic, nine leafy vegetables using salty water to simulate drainage water, where the substrate bags are filled with perlite with a pH of 7, finding that swiss chard was the most salt tolerant among those an electrical conductivity of 0.09 dS·m−1 , and a granulometry evaluated. With a similar method, Grieve and Suarez (1997) of [0–6]. focused on potential sulfate and selenium toxicity, concluding The RTG section used for the donor tomato crop was that purslane (Portulaca oleracea) is an excellent candidate for southeasterly facing and occupied an area of 84.3 m2 . 57 substrate saline drainage water reuse systems. Real case studies with bags were used (40L), planting 3 seedlings per bag, totaling cascade systems were analyzed by Muñoz et al. (2017), who 171 plants. The tomato crop was planted on January 14th and assessed the performance of beef tomato in a greenhouse as the uprooted on August 2nd, 2019. The average concentration of donor crop and a sequence of lettuce, tomato, and endive as fertilizers in kg/m3 was KPO4 H2 – 0.283, KNO3 – 0.138, K2 SO4 – receiving crops. This study underlined the need to further study 0.367, Ca(NO3 )2 – 0.533, CaCl2 H2 O – 0.133, Mg(NO3 )2 – 0.178, the observed yield decrease and its causes. Incrocci et al. (2003) Hortilon – 0.011, and Sequestrene – 0.011, although the nutrient assessed a cascade cropping system using tomato as the donor solution was adapted based on the evolution and phenological crop and cherry tomato as the receiving crop, highlighting the stages of the tomato crop. The leachates of the tomato crop potential of cascade set-ups to increase the water use efficiency were collected through connected pipes from leachate trays and of the system. Nonetheless, little is known about the potential transported to a 300-L tank using a submergible pump. of a cascade system to diminish the nutrient load of drainage To maximize the nutrient recycling, these leachates were flows. García-Caparrós et al. (2016) explored this potential with used to irrigate two crops. First, the same tomato crop using a pot experiment using ornamental Juncus acutus. The authors a recirculation system (totaling 7.5 m3 ), whose performance found that the species used presented a good bioremediation was left out of the study to focus only on the cascade system. potential based on the uptake of nitrogen (N) and phosphorus Second, a receiving lettuce crop located in another section of the (P). García-Caparrós et al. (2018) also explored this potential RTG southwesterly facing using 2 L/h drippers (totaling 8.8 m3 ). using horticultural melon as the donor crop and ornamental Figure 1 shows the diagram of the water flows of the system. rosemary as the receiving crop, finding a decrease in yield in Five successive lettuce cycles were grown as shown in Table 1. the receiving crop, but also a great potential to optimize the The choice of lettuce was based on the lower nutritional demand water flows and remove the nitrates of the drained water. To our from this crop compared to tomato. Low nutrient concentrations knowledge, there are no references in the literature exploring the were expected in the leachates of the tomato crop, and thus lettuce potential of horticultural species (both as donor and receiving can potentially produce competitive yields with the nutrients Frontiers in Plant Science | www.frontiersin.org 2 November 2020 | Volume 11 | Article 596550 Rufí-Salís et al. Cascade System in Urban Agriculture FIGURE 1 | Diagram of the cascade system. supplied through the cascade system. Average temperatures the concentration in the water flows, the respective samples and relative humidity per test are shown in Supplementary were collected three times per week and analyzed using ion Figures S1, S2, respectively. chromatography (ICS-1000 and AS-DV by Dionex) to obtain Plants were distributed in 4 substrate trays, with 8 perlite bags concentration results for the anions nitrite (NO2 − ) and nitrate each using a 4 × 2 organization and with a 1.2-m separation (NO3 2− ). The same samples were externally analyzed using between trays. Each substrate bag can allocate 4 plants, totaling 32 ICP-OES atomic spectroscopy (Optima 4300DV by Perkin- plants per tray. All campaigns were planted from nursery plants. Elmer) to obtain results for P, potassium (K), magnesium (Mg), Besides campaigns three to five of the cascade system, calcium (Ca), and sulfur (S). Nutrient content in the lettuce we planted mirror control crops using the same species and was determined based on three plants randomly selected at variety to compare the results in terms of yield and nutrient the end of the harvest. Samples were sorted into envelopes content during the seasonal period with expected higher water and dried in a furnace at 65◦ C for 1 week before analyzing demand and salinity. We used 2 substrate trays for the control, externally the concentration of P, K, Ca, Mg, and S via ICP-OES totaling 64 plants. atomic spectroscopy (Optima 4300 DV by Perkin Elmer) and the concentration of N with elemental analysis (Flash EA 2000 CHNS Water and Nutrient Monitoring by Thermo Fisher Scientific). Since the sample for every test and To assess the water flows in the cascade system, analogic treatment is 3 random plants analyzed twice, a formal statistical flow meters were coupled to the water system. Samples of the analysis was not performed due to the low sample size. irrigation in the receiving crop and the control were collected To calculate the accumulated irrigated nutrients per test, the directly from the drippers placed in the perlite bags. To determine concentration values obtained through the continuous analysis of the water samples for each nutrient were multiplied by the amount of water irrigated between the date a specific sample was taken and the date of the previous sample. The values obtained TABLE 1 | Number of plants and calendar of tests undergone in the present study. were then aggregated to obtain the final accumulated irrigated Test Treatment Planted Harvested Plant number nutrients per test, and divided by the number of plants in a specific test to obtain this parameter per plant. T1 Cascade February 8 March 5 128 T2 Cascade March 5 April 11 128 T3 Cascade April 24 May 31 128 Statistical Methods Control 64 A set of statistical methods and tests were used to verify and T4 Cascade May 31 July 1 128 strengthen the results obtained in different sections. The tests Control 64 were performed through the use of different packages developed T5 Cascade July 1 August 1 128 for R programming software, considering a statistical significance Control 64 when p-values were less than 0.05. Frontiers in Plant Science | www.frontiersin.org 3 November 2020 | Volume 11 | Article 596550 Rufí-Salís et al. Cascade System in Urban Agriculture Shapiro Wilk’s test (Shapiro and Wilk, 1965) was used as the The third method consists on the analysis of the “Estimated method to analyze the normality of observations or residuals Marginal Means Of Linear Trends” through the “emtrends” obtained through linear regressions. Levene’s test (Levene, 1961) function from the “emmeans” R package (Lenth et al., 2020). As was used as the method to analyze the homogeneity of variances. stated by Lenth et al. (2020), “emtrends is useful when a fitted Levene’s test is an alternative to Bartlett’s test with less sensitivity model involves a numerical predictor interacting with another to non-normal data. If the two previous tests (Shapiro-Wilk’s predictor.” and Levene’s) presented p > 0.05, we applied an analysis of covariance (ANCOVA) to the linear regression to evaluate if the means of a dependent variable were equal under different levels of RESULTS AND DISCUSSION an independent variable. Oppositely, we applied Kruskal-Wallis test (Kruskal and Wallis, 1952) as the non-parametric analog This section presents and discusses the results divided in of ANCOVA, recommended with non-normal distributions and different sections: see section “Production,” section “Water: inequality of variances (Feir-Walsh and Toothaker, 1974) to Irrigation,” section “Water: EC and pH,” section “Water: check if samples came from the same distribution. Nutrient Content,” section “Biomass: Nutrient Content,” section Three different methods were used to compare the statistical “Decreasing Eutrophication Potential and Nutrient Depletion, significance between slopes of a linear regression. The first and section “Scaling of the System.” Values in X ± Y form express one was based on Meseguer-Lloret (2017b) for homogeneous average ± standard deviation. variances and Meseguer-Lloret (2017a) for non-homogeneous variances, and is based on a t-test. This method requires a Production prior test for normality. If a normal distribution is observed, a Figure 2 shows the average fresh weight per lettuce in every test. F-test can be performed to determine if the variances between We can see that Tests 1 (February) and 3 (May), produced low treatments were homogeneous, since it is accurate only for and similar yields: 108.66 ± 26.85 and 109.82 ± 22.39 g/plant, normal data. However previous research concluded that the t-test respectively. Test 2 (April) produced slightly more yield with is highly robust for equal sample size (Welch, 1938; Milton and a higher variability of data (127.95 ± 34.60 g/plant), while Tsokos, 1983), which is the case for most data in this study. Test 4 (June) and Test 5 (July) exerted the highest yields, with Based on the method described by Meseguer-Lloret (2017a) for 220.31 ± 38.27 and 134.17 ± 34.84 g/plant, respectively. On non-homogeneous variances, a t-test should be performed by the other hand, the control treatments in Test 3, 4, and 5 calculating tcalc as expressed in Eq. 1 and t’calc as expressed in Eqs (254.91 ± 52.66, 232.49 ± 62.25, and 185.11 ± 32.75 g/plant) 2–4, being “b” the value of the slope, “s2 ” the variance, and “t” the always had more weight than its respective cascade crops. specific value in the t-table of probabilities. The low yields obtained in the receiving crop in the cascade system coincide with the findings by Muñoz et al. (2012), who |b1 − b2 | tcalc = q (1) used tomato as both donor and receiving crop and Muñoz s2b1 + s2b2 et al. (2017), who used tomato as donor crop and lettuce as receiving crop. 0 t1 · s2b1 + t2 · s2b2 We determined that only cascade’s Test 1 (p = 0.78) and 4 tcalc = (2) (p = 0.26) and control’s Test 5 (p = 0.83) presented a normal s2b1 + s2b2 distribution (p > 0.05). However, a normal distribution was n1 −2 observed when the test was done including all the control t1 = t95% (3) data (p = 0.22), but not when analyzing the entire cascade dataset (p < 0.05). Both of these results were expected based n2 −2 t2 = t95% (4) on the homogenous irrigation in the control and the foreseeable variability in nutrient content of the leachates of the donor If tcalc > t’calc , the slopes analyzed are statistically different. On the crop, analyzed in-depth in section “Water: Nutrient Content”. other hand, based on the method described by Meseguer-Lloret When normality is studied for the residuals obtained through (2017b) for homogeneous variances, a t-test should be performed multiple combinations between tests and treatments, a non- to calculate tcalc as expressed in Eq. 1 and ttab , which is the value normal distribution is observed in all of them (p < 0.05). On in the t-table with a 95% of probability and “n1 + n2 -2” degrees the other hand, the biggest differences between treatments were of freedom, as expressed in Eq. 5. found in Test 3 and Test 5, in which the control produced 132 and n2 −2 ttab 95% = t95% (5) 37% more yield, respectively, than their simultaneous cascade tests. Oppositely, the differences between treatments in Test 4 If tcalc > ttab , the slopes analyzed are statistically different. were narrowed to 6%. Despite this big variability, none of the Oppositely, the slopes analyzed are statistically similar if comparisons between treatments (i.e., Control vs. Cascade in Test tcalc < ttab . 3, 4, and 5) presented a significant homogeneity of variances The second method is based on Andrade and Estévez-Pérez (p < 0.05). Based on the non-normality detected, we applied (2014). Although the authors state that the t-test is a robust Kruskal-Wallis test to compare between treatments. As expected, method to compare the statistical difference between slopes, they the results showed that the distribution differences between suggest the use of the ANCOVA as a simpler method. cascade and control were significant in every test (p < 0.05). Frontiers in Plant Science | www.frontiersin.org 4 November 2020 | Volume 11 | Article 596550 Rufí-Salís et al. Cascade System in Urban Agriculture FIGURE 2 | Production of the different tests. FIGURE 3 | Irrigation per plant of the different tests. Water: Irrigation cascade system and the control in Test 3. Control Test 3 irrigated All water irrigated to the receiving crop (8.8 m3 ) was supplied 20.30 L/plant with a slope of 0.60 (R2 = 1.00), doubling the slope through the leachates of the donor crop, thus being a residual flow of cascade Test 3 (0.24 with R2 = 0.99), that irrigated 9.60 L/plant. that the lettuce crop is taking benefit from and avoiding the use The statistical differences between the slopes obtained in the of new water. Irrigation was increased in every successive crop different tests and treatments were determined through a multi- based on the climatic conditions inside the greenhouse, irrigating step method with a final t-test. First, we determined if the 5 times more water in Test 5 than in Test 1 (Figure 3). The individual accumulated irrigation per plant, test and treatment slope of the linear regression in Test 1 and 2 of the cascade followed a normal distribution, which was confirmed for all of treatment was similar (0.22 and 0.23 with R2 of 0.96 and 1.00, them (p > 0.05). A normal distribution was also confirmed for respectively). This minimum difference was expected considering the residuals (p > 0.05). In the next step, the variances between the temperatures reached inside the greenhouse during Test 1 and treatments were labeled as non-homogeneous in Test 3 (p < 0.05) 2 periods, which were really similar (Supplementary Figure S1). and homogeneous in Test 4 and 5 (p > 0.05). Considering the We observed an anomaly in the water irrigated between the non-homogenous variances in Test 3 we calculated tcalc and t’calc , Frontiers in Plant Science | www.frontiersin.org 5 November 2020 | Volume 11 | Article 596550 Rufí-Salís et al. Cascade System in Urban Agriculture obtaining the values 25.85 and 2.22, respectively. Considering Water: EC and pH that tcalc > t’calc , we concluded that the slopes between cascade Figure 4 shows the electrical conductivity (EC) values for the and control in Test 3 were statistically different. On the other entire period under study. Since the control irrigation was hand, we performed a t-test for homogeneous variances for Test manually controlled through the addition of fertilizers based 4 and 5. ttab values for Test 4 and 5 were both 1.71. tcalc for Test on crop requirements, the EC values showed a great stability. 4 was 6.38, while a value of 0.14 was obtained for Test 5. Given On the other hand, the EC of the water used to irrigate the that tcalc > ttab for Test 4, the slopes between treatments in this cascade crops increased over time, as the leachates of the tomato test were observed to be statistically different. Oppositely, the crop got richer in nutrients. The highest value was reached slopes between treatments in Test 5 were statistically similar since in the final cascade cycle, with 2.78 dS/m, much higher than tcalc < ttab . the irrigation of the control (1.46 dS/m). Abou-Hadid et al. The ANCOVA applied to the data in Figure 3 strengthened the (1996) states that the perfect EC value for growing lettuce falls statistical results obtained in the t-test since the same conclusions within the 1.0–1.5 dS/m range. Other literature specific for were obtained: differences between slopes in Test 3 and 4 were hydroponic lettuce mention that slightly higher values (1.2– found to be statistically different, while in Test 5 were observed 1.8 dS/m) could be beneficial (Singh and Dunn, 2016). Most to be statistically similar. experiments found in the literature report greater yields within To double-verify the statistical conclusions, we analyzed these values (Serio and Elia, 2001; Miceli et al., 2003; Samarakoon the “Estimated Marginal Means Of Linear Trends.” The et al., 2019). Among the cascade test performed in our study, outcome of the analysis was the same as the one obtained Test 1 (0.92 dS/m), Test 2 (1.25 dS/m) and Test 3 (1.68 dS/m) through the t-test and ANCOVA: statistically differences fell within the optimal range reported by Singh and Dunn (2016) between slopes in Test 3 and 4 (p < 0.05) and statistically (Figure 4 – highlighted area). EC readings for drainage water in similar slopes for Test 5 (p = 0.89) between cascade and cascade systems are higher than the ones observed in Figure 1, control treatments. such as 3.62 dS/m reported by García-Caparrós et al. (2016) in The difference in irrigation rate in Test 3 was related to a drained water of Ruscus aculeatus. The authors also reported shortening in the amount of leachates generated by the tomato that a dilution 1:2 with fresh water was able to decrease the crop, coupled with the parallel irrigation with leachates in the EC to values <3 dS/m. In this sense, to compensate the high donor crop. To solve this problem, three options were considered. EC reached in the final cascade cycle, we added fresh water in First, to increase the irrigation in the donor crop, thus leaching the leachates, thus triggering the unstable values observed in more water that would be available for the receiving crop. This Figure 3 for Test 5. However, EC values obtained in the present option was discarded because increasing the irrigation water use study weren’t as unsuitable as expected considering the focus efficiency (assumed to be the volume of irrigated water applied on salt tolerance from the literature around cascade systems. per unit of yield) of the receiving crop while decreasing this Oppositely to EC, pH values decreased overtime for the cascade parameter for the donor crop seemed contradictory. Moreover, cycle, as shown in Supplementary Figure S3. A pH optimal the composition of the leachates will likely vary if more water range to grow hydroponic lettuce of 6.0–7.0 is reported by Singh is used in the donor crop. Second, to add water to the cascade and Dunn (2016), and of 5.6–6.0 by Brechner and Both (2013). system to meet the demand of the receiving crop. This option Considering a broad range, all cascade tests presented optimal was also discarded because it would imply a modification in pH mean values except Test 1, that had a mean pH value the composition of the irrigation of the receiving crop, thus slightly higher than 7. blurring the effect of a cascade system. More important, both of these strategies were contradictory with the general aim of cascade systems: to take advantage of the residual flows of water Water: Nutrient Content and nutrients. If changes in the predefined system are needed, Figure 5 shows the accumulated irrigated nutrients per plant per the value of the implementation of cascade systems decrease test. As expected, the nutrients irrigated in the cascade treatment drastically. What could be seen as the best strategy is based on the increased over time, coinciding with the amount of nutrients practitioner or farmer priorities. When there is shortage of water, leached by the tomato crop. K was the nutrient with the highest the farmer should prioritize in which system the residual flows are quantities in the cascade irrigation (reaching 10.7 g/plant in used. With a cascade set-up, the receiving crop can be prioritized Test 5 of the cascade system), followed by Ca and N, reaching during water shortage periods, minimizing the recirculation of 5.0 and 3.6 g/plant in Test 5, respectively. On the other hand, water in the donor crop itself. However, if the donor crop is nutrients irrigated in the control treatment presented a similar already not recirculating its own water and there is still water behavior among tests. shortage, the first and second options outlined in this section can Although the linear regressions observed in section “Water: be applied considering the mentioned limitations. Irrigation” regarding the irrigated water presented R2 values The differences between treatments in Test 4 and 5 were around 0.99, the evolution of the accumulated irrigated nutrients narrower than the ones observed in Test 3 since there were not per plant (Figure 5) could also be affected by the concentration episodes of water shortening. Control Test 4 presented a slope of nutrients in the water flows. As expected, the nutrient in the of 0.82 (R2 = 0.99), 18% higher than the slope for cascade Test control irrigation followed a completely linear regression in all 4 (0.69; R2 = 1.00). Slope for Test 5 was the same for both nutrients (R2 ≈ 0.99) since the nutrient solution supplied to the treatments (0.81; R2 = 0.99). crop is completely homogenous. On the other hand, the linear Frontiers in Plant Science | www.frontiersin.org 6 November 2020 | Volume 11 | Article 596550 Rufí-Salís et al. Cascade System in Urban Agriculture FIGURE 4 | Electrical conductivity (EC). Highlighted area represents the suitable EC range to grow hydroponic lettuce stated by Singh and Dunn (2016). FIGURE 5 | Accumulated irrigated nutrients per plant per test. regression of the accumulated irrigated nutrients in the cascade its stability can be highlighted as a relevant parameter to also treatment highly fitted with the observations, with R2 values stabilize the nutrient content of the leachates and thus, provide higher than 0.97. The only exception was N in Test 4, which a balanced and reliable source of nutrients to the receiving crop presented a R2 = 0.85. The reason behind this behavior was the in a cascade system. low quantities of this nutrient available through the leachates The slope of the accumulated irrigated nutrient per plant per of the donor crop in Tests 1, 2, and 3. In this sense, we can test was compared between treatments for every specific nutrient see that this N limitation is overcome in the middle of Test 4 through ANCOVA (Andrade and Estévez-Pérez, 2014). A prior (concentration jumps from 35 to 111 mg/L), probably related step was done to assess the normality of the residuals, confirming to a N increase in the nutrient solution supplied to the donor the normality for all nutrients and tests apart from N in Test 3 crop. In this sense, the nutrient supply to the donor crop and with a p < 0.05. An analysis to determine the homogeneity of Frontiers in Plant Science | www.frontiersin.org 7 November 2020 | Volume 11 | Article 596550 Rufí-Salís et al. Cascade System in Urban Agriculture variances was applied to all combinations of nutrients and tests. irrigated through the cascade system (Supplementary Table S2). As expected, Test 3 for N also presented statistically different Supplementary Table S3 shows the amount of nutrients leached variances (p < 0.05). Nonetheless, other cases were detected: P, by the lettuce plants. The ratio between nutrients taken up and K Ca and Mg for Test 3, S for Test 4 and K, Mg and S for Test irrigated decreased over time given the fact that the amount of 5. In this sense, ANCOVA was applied to the remaining cases nutrients being irrigated through the cascade system increased with normal distribution and homogeneity of variances: Test 3 as the leachates from the tomato crop got richer. We can see for S, Test 4 for N, P, K, Ca and Mg and Test 5 for N, P, and Ca, that all N irrigated through the cascade system was taken up by observing statistically different slopes between treatments in all of the lettuce crop in tests 1 and 2, with two major consequences. them (p < 0.05). The same outcome was obtained when analyzing First, a possible nitrogen deficiency that made Test 1 the one with the “Estimated Marginal Means Of Linear Trends” for these the lowest yield, making the cascade set-up inefficient in terms cases, although one exception was detected. P in Test 4 presented of production. Second and with a bioremediation perspective, statistical different slopes through ANCOVA (p < 0.05), but the removal of all N from the residual flow used to irrigate the statistical similarity was observed through “emtrends.” receiving crop triggered the complete mitigation of the marine eutrophication impacts caused by the leachates of the donor crop. Biomass: Nutrient Content This mitigation decreased over time, getting down to 43.1% of Figure 6 shows the nutrient content in the lettuce for cascade removal in Test 3, and being only 6.5% in Test 5. The removal and control tests. Every nutrient had a different behavior among of P was already below one third in Test 1, thus just mitigating tests, especially in the cascade treatment. N increased over time, part of the freshwater eutrophication impacts. The removal rate especially between the 2nd and the 3rd test, and the 3rd and of P got below 10% in Test 2, and it reached its lowest level in the 4th test. The 5th test presented the highest concentration of Test 5 with only 1.9%. Given the non-renewable nature of P, the this nutrient. This tendency was expected based on the findings reuse of P is not only relevant to avoid freshwater eutrophication, of the previous section, where we observed the increasing but also in terms of mitigating its depletion. K was the nutrient concentration of N in the leachates of the donor crop. Although irrigated with the most quantity through the cascade system the tendency related to the concentration in the water flow (Figure 4). However, high removal rates were only reached in was observed for most nutrients, the 5th test was not the one Test 1 (63.4%). Removal rate in Test 2 already got below 25% and with the highest nutrient content in the biomass for other kept decreasing over time. nutrients analyzed. The 4th test was the one with the highest Since the irrigation in the control was more controlled both in concentrations for P, K, Mg, and S with 0.97, 11.43, 0.34, and terms of water and nutrients, the removal rates among tests were 0.35%, respectively. For Ca, the concentration in the 1st test was more stable. This was specially the case of N, Ca, and Mg, while higher than the one observed for the 4th test (1.65%), which P slightly decreased in control tests 4 and 5, probably due to a doesn’t correlate with its supply through the cascade system combination of changes in the nutrient solution and an increase (Figure 5), and neither with the concentration of Ca in the in P consumption of the donor crop in the production stage. leachates flow from the donor crop, which increased overtime. This high absorption of Ca in the first test neither correlates with Scaling the System traditional compatibilities/incompatibilities between nutrients Other studies that considered the use of a the leachates of a (Mulder, 1956), since absorption of Ca may be depressed by donor crop to irrigate horticultural products did not account excessive amount of Mg, which was also highly absorbed in Test for the variability of nutrients along the cycle of the donor 1, or favored with excessive nitrate concentrations, which was crop. For example, Choi et al. (2011) first collected the waste limitedly supplied through the cascade system in the first tests. nutrient solution from a hydroponic tomato crop and then As expected, the control presented less differences between applied it to a cabbage crop instead of connecting a donor and tests, especially for N. Due to differences in irrigated water a receiving crop simultaneously in a cascade system. Therefore, between the cascade and the control in Test 3, nutrient content in the nutrient concentration supplied by the authors was stable, the control was higher than in the cascade treatment for this test avoiding problems related to variability but still subjected to for all nutrients. When the irrigated water differences between possible nutrient limitations. However, NO3 − concentration in the control and the cascade tests was normalized again in Tests Choi et al. (2011) study (1285 mg/L) was always above all the 4 and 5, differences between treatments decreased. The average NO3 − values determined in our study. Considering the higher N content for the control (4.4%) was really similar to the one nutrient content of lettuce compared to cabbage (Weber, 2016), presented in tests 4 and 5 of the cascade system (4.4%). We could it is not surprising that Choi et al. (2011) obtained higher also find really similar concentrations between treatments in Test yields than the control. In a similar experiment, Zhang et al. 4 for K (≈11.02%) and S (≈0.35%), and in Test 5 for P (≈0.78%), (2006) also obtained higher yields using tomato leachates in K (≈9.9%) and Ca (≈1.3%). muskmelon and cucumber. By using the same donor crop, Incrocci et al. (2003) irrigated a cherry tomato cycle. Since the focus of the author was on the effect of salinity, the waste Decreasing Eutrophication Potential and nutrient solution was adjusted in terms of nutrient content to Nutrient Depletion meet the same amount as the control: 161 mg/L of N. This Table 2 shows the amount of nutrients that were taken up by concentration was lower than the one used by Choi et al. (2011). the lettuce plants (Supplementary Table S1) with respect to the However, this value was not reached in the cascade system of Frontiers in Plant Science | www.frontiersin.org 8 November 2020 | Volume 11 | Article 596550 Rufí-Salís et al. Cascade System in Urban Agriculture FIGURE 6 | Nutrient context per plant (%). the present study until Test 4, in which the yield differences crop of the present study (72.5 mg/L). This probably indicates between the control and the cascade plants was the narrowest excessive N in the irrigation of the donor crop, which could be (Figure 2). With a cascade set-up using tomato as both donor adjusted without yield reductions as suggested and quantified by and receiving crops, Muñoz et al. (2012) converge with the the same authors. present study since the authors observed yield reductions in However, none of these authors considered the variability the receiving crop compared to the control [as also found by of nutrient content as a relevant parameter in cascade systems. Muñoz et al. (2017)], irrigated with N concentrations of 198 Taking into account the need for a suitable dimensioning between and 310 mg/L, respectively. The N concentration in the tomato donor and receiving crops in cascade systems to define best leachates in Muñoz et al. (2012) (198 mg/L) is more than two practices (Muñoz et al., 2017), we scaled the system in terms of times the average concentration leached by the tomato donor nutritional demand of the receiving crop. Frontiers in Plant Science | www.frontiersin.org 9 November 2020 | Volume 11 | Article 596550 Rufí-Salís et al. Cascade System in Urban Agriculture TABLE 2 | Uptake (%) of irrigated nutrient per plant in cascade treatments. Nutrients T1 – Cas (%) T2 – Cas (%) T3 – Cas (%) T3 – Con (%) T4 – Cas (%) T4 – Con (%) T5 – Cas (%) T5 – Con (%) N 143.8* 115.9* 43.1 17.1 23.3 16.4 6.5 16.7 P 27.8 9.0 8.7 10.1 6.4 6.9 1.9 5.0 K 63.4 24.4 17.4 31.8 13.4 19.8 5.1 16.0 Ca 23.7 6.9 5.4 4.6 5.1 4.2 1.4 4.0 Mg 15.2 6.4 6.3 8.1 6.3 7.3 2.4 8.1 S 4.5 1.6 1.1 3.2 1.3 3.0 0.6 2.0 Cas, cascade system; Con, control. *Values above 100% mean complete uptake. Due to the small amount of nutrients leached by the donor crop during T1 and T2 the experimental and analytical error is increased. TABLE 3 | Amount of lettuce that could be produced in the cascade system under analysis considering the application of all tomato leachates. Units of lettuce plants/tomato plant Units of lettuce plants/171 tomato plants Nutrients T1 T2 T3 T4 T5 T1 T2 T3 T4 T5 N 0.1 0.6 1.7 6.3 9.1 17.1 101.8 286.4 1073.5 1554.2 P 0.9 8.5 7.1 29.6 30.0 145.6 1460.9 1216.5 5065.1 5131.1 K 0.4 3.4 4.5 12.8 11.0 63.9 585.3 769.7 2188.8 1874.6 Ca 1.6 12.2 13.6 36.8 40.3 267.6 2091.5 2329.4 6288.4 6888.7 Mg 2.1 13.3 12.8 29.5 28.7 366.6 2275.7 2181.9 5036.6 4912.8 S 5.2 52.9 65.3 139.1 99.4 897.6 9042.9 11161.7 23779.4 17005.9 To quantify the scaling between donor and receiving crops irrigated using a conventional nutrient solution. With this option, in the system under study, we considered the average uptake of yield would be maximized, while adapting the nutrient demand the control for all nutrients [N (4.4%), P (0.84%), K (11.0%), Ca of the lettuce cycles to the nutrient supplied through the cascade (1.4%), Mg (0.29%), and S (0.33%)] and the average yield of the system without requiring further nutrient management. Third control (224 ± 58 g/plant). Table 3 shows the quantity of lettuce and finally, start with 1554 lettuce plants connected to the cascade that could be grown with the maximum quantity of nutrients that system. To avoid nutrient deficiencies as the ones observed in the could be supplied through the cascade system, considering that present study, the nutrient content in the leachates tank used to all leachates of the donor crop are applied to the receiving crop irrigate the receiving crop must be adjusted. A specific amount of through the cascade system. all nutrients should be added in Test 1, while N, P and K should As we can see in Table 3, the limiting nutrient was always be added in Test 2 and 3, and only N should be added in Test N, allowing to grow a 1:10 and a 6:10 system for Test 1 and 4. This option would be the most efficient to produce the same 2, respectively. P (9:10) and K (4:10) in Test 1 were the other amount of lettuce over time while minimizing the nutrient input. nutrients with less than a 1:1 ratio between the tomato donor However, it would require a detailed control of nutrient flows to crop and the lettuce receiving crop. Considering N as the limiting avoid nutrient deficiencies. It is important to understand that, despite the strategies mentioned above, the list of management nutrient, we can state that in our system with 171 tomato plants, practices is endless: mix and store the leachates during Test 1 to we could grow 17 lettuce plants in Test 1, while being able to increase the nutrient supply to Test 2 (and omit Test 1 of lettuce), grow 1554 lettuce plants in Test 5, thus giving the practitioners increase the water irrigated to the donor crop, etc. However, the three main options in terms of crop management, considering value of all possible strategies rely on their performance in real an ideal situation where we have an area that can allocate 1554 crops. Therefore, we highlight the need to evaluate and quantify lettuce plants, which is the number of lettuces that we could grow the potential yield that could be reached through the application in Test 5 considering the N content in the tomato leachates. of these strategies. In addition, the analysis of the nutritional and First, to grow 17 lettuce plants in all tests. This option will water flows in pilot or full-scale experiments should be included ensure the production of the same amount of plants in every in the list of upcoming challenges to verify that the plant nutrition test but will present huge inefficiencies in terms of space (if the is optimal and that the cascade system is well scaled to mitigate rest of the area is not occupied by other crops) and nutrient eutrophication impacts derived from the depletion of nutrients flows, discharging a great quantity of nutrients that will increase in the donor crop. over time as the leachates of the donor crop become nutrient- richer. This nutrient inefficiency could be mitigated by applying recirculation in the donor crop itself. Second, to grow 17 (Test 1), CONCLUSION 101 (Test 2), 286 (Test 3), 1073 (Test 4), and 1554 (Test 5) lettuce plants connected to the cascade system, while growing 1537 (Test The present paper has presented an evaluation of a cascade 1), 1453 (Test 2), 1268 (Test 3), and 481 (Test 4) lettuce plants system with a long-cycle tomato donor crop and five Frontiers in Plant Science | www.frontiersin.org 10 November 2020 | Volume 11 | Article 596550 Rufí-Salís et al. Cascade System in Urban Agriculture successive cycles of lettuce. The assessment of the agronomic AUTHOR CONTRIBUTIONS performance and the nutrient flows have shed light on the potential of these systems to mitigate nutrient depletion MR-S, AP-B, GV, and XG conceived the original idea for the in cities while producing food in the framework of study. MR-S, VA-P, and FP set up, supervised, and acquired the urban agriculture. data for the experimental tests. MR-S processed and analyzed the The variation of nutrient content of the leachates produced data and took the lead in writing the manuscript. All authors were by the donor crop is a key parameter to plan the amount of responsible for the conception and design of the study, critically plants that can be planted of the receiving crop. The early stage revised the draft for important intellectual content, and gave their of the donor crop could only produce 0.1 lettuces per tomato final approval to the manuscript. plant, with N as the limiting nutrient. On the other hand, the late stage of the donor crop was able to leach enough nutrients to feed 9 lettuces per tomato plant. However, attention must FUNDING be paid on the electrical conductivity of the water flow to stay within non-harmful values. Nevertheless, the cascade system was The authors are grateful to Universitat Autònoma de Barcelona shown to be efficient to mitigate the nutrient discharge of open for awarding a research scholarship to MR-S (PIF-UAB 2017), systems, especially in terms of N and P to avoid eutrophication the Spanish Ministry of Economy, Industry and Competitiveness impacts in the early stage of the tomato crop. To this end, a good (Spain) for the grant awarded to VA-P (FPI-MINECO 2018), scaling between the two crops of the system is vital to tap the full and the National Commission for Scientific and Technological potential of the cascade set-up, while having different options in Research (Chile) for the grant awarded to FP (PFCHA- terms of system management. CONICYT 2018 – Folio 72180248). The research leading to this Given the findings of this study, we encourage future publication has received funding from the European Union’s researchers to test different kind of horticultural crops. Horizon 2020 Research and Innovation Program under grant Considering the nutritional problems in the beginning of the agreement number (FoodE project) 862663. The publication cycle of the donor crop and the harmful salinity that can reflects the author’s views. The Research Executive Agency be reached at the end, further research should test possible (REA) is not liable for any use that may be made of the combinations of donor and receiving crops that minimize information contained therein. This work was supported by the these two problems. Reporting the limitations of these kind Spanish Ministry of Economy, Industry and Competitiveness of systems is key to a transparent process of decision- (AEU/FEDER) [CTM2016-75772-C3-1-R] and the “María de making in the implementation of optimization strategies in Maeztu” program for Units of Excellence in R&D [CEX2019- urban agricultural systems. In terms of experimental design, 000940-M]. AP-B thanks the German Federal Ministry of further research assessing the nutritional flows of cascade Education and Research for the financial support of the research systems should increase the number of plants that will group “Circulus – Opportunities and challenges of transition to a be analyzed in terms of nutrients to precisely determine sustainable circular bio-economy,” grant number 031B0018. This the variability of concentrations within the same treatment work was supported by and ERC Consolidator grant awarded to and test. Gara Villalba (818002-URBAG). SUPPLEMENTARY MATERIAL DATA AVAILABILITY STATEMENT The Supplementary Material for this article can be found The raw data supporting the conclusions of this article will be online at: https://www.frontiersin.org/articles/10.3389/fpls.2020. made available by the authors, without undue reservation. 596550/full#supplementary-material REFERENCES Bocken, N. M. P., Olivetti, E. A., Cullen, J. M., Potting, J., and Lifset, R. (2017). Taking the circularity to the next level: a special issue on the circular economy. Abou-Hadid, A. F., Abd-Elmoniem, E. M., El-Shinawy, M. Z., and Abou-Elsoud, J. Ind. Ecol. 21, 476–482. doi: 10.1111/jiec.12606 M. (1996). Electrical conductivity effect on growth and mineral composition Boneta, A., Rufí-Salís, M., Ercilla-Montserrat, M., Gabarrell, X., and Rieradevall, of lettuce plants in hydroponic system. Acta Hortic. 434, 59–66. doi: 10.17660/ J. (2019). Agronomic and environmental assessment of a polyculture rooftop ActaHortic.1996.434.6 soilless urban home garden in a mediterranean city. Front. Plant Sci. 10:341. Agung Putra, P., and Yuliando, H. (2015). Soilless culture system to support water doi: 10.3389/fpls.2019.00341 use efficiency and product quality: a review. Agric. Agric. Sci. Proc. 3, 283–288. Brechner, M., and Both, A. J. (2013). Hydroponic Lettuce Handbook. Kentucky: doi: 10.1016/j.aaspro.2015.01.054 University of Kentucky. Ahmed, A. K., Cresswell, G. C., and Haigh, A. M. (2000). Comparison of sub- Carmassi, G., Incrocci, L., Maggini, R., Malorgio, F., Tognoni, F., and Pardossi, irrigation and overhead irrigation of tomato and lettuce seedlings. J. Hortic. Sci. A. (2005). Modeling salinity build-up in recirculating nutrient solution culture. Biotechnol. 75, 350–354. doi: 10.1080/14620316.2000.11511249 J. Plant Nutr. 28, 431–445. doi: 10.1081/PLN-200049163 Andrade, J. M., and Estévez-Pérez, M. G. (2014). Statistical comparison of the Choi, B.-S., Lee, S.-S., and Ok, Y.-S. (2011). Effects of waste nutrient solution on slopes of two regression lines: a tutorial. Anal. Chim. Acta 838, 1–12. doi: growth of chinese cabbage (Brassica campestris L.) in Korea. Korean J. Environ. 10.1016/j.aca.2014.04.057 Agric. 30, 125–131. doi: 10.5338/kjea.2011.30.2.125 Frontiers in Plant Science | www.frontiersin.org 11 November 2020 | Volume 11 | Article 596550 Rufí-Salís et al. Cascade System in Urban Agriculture De Pascale, S., and Maggio, A. (2005). Sustainable protected cultivation at a Muñoz, P., Paranjpe, A., Montero, J. I., and Antón, A. (2012). Cascade crops: an mediterranean climate. Perspectives and challenges. Acta Hortic. 691, 29–42. alternative solution for increasing sustainability of greenhouse tomato crops in doi: 10.17660/ActaHortic.2005.691.1 mediterranean zone. Acta Hortic. 927, 801–805. doi: 10.17660/actahortic.2012. Ellen MacArthur Foundation and McKinsey Center for Business and Environment 927.99 (2015). Growth Within: A Circular Economy Vision for a Competitive Europe. Romero-Gámez, M., Audsley, E., and Suárez-Rey, E. M. (2014). Life cycle Cowes: EMF. assessment of cultivating lettuce and escarole in Spain. J. Clean. Prod. 73, Feir-Walsh, B. J., and Toothaker, L. E. (1974). An empirical comparison 193–203. doi: 10.1016/J.JCLEPRO.2013.10.053 of the anova f-test, normal scores test and kruskal-wallis test under Romero-Gámez, M., Suárez-Rey, E. M. M., Antón, A., Castilla, N., and Soriano, T. violation of assumptions. Educ. Physchol. Meas. 34, 789–799. doi: 10.1177/ (2012). Environmental impact of screenhouse and open-field cultivation using 001316447403400406 a life cycle analysis: the case study of green bean production. J. Clean. Prod. 28, García-Caparrós, P., Llanderal, A., El-Tarawy, A., Correia, P. J., Pestana, M., and 63–69. doi: 10.1016/j.jclepro.2011.07.006 Lao, M. T. (2016). Irrigation with drainage solutions improves the growth and Rufí-Salís, M., Petit-Boix, A., Villalba, G., Sanjuan-Delmás, D., Parada, F., Ercilla- nutrients uptake in Juncus acutus. Ecol. Eng. 95, 237–244. doi: 10.1016/j.ecoleng. Montserrat, M., et al. (2020). Recirculating water and nutrients in urban 2016.06.090 agriculture: an opportunity towards environmental sustainability and water use García-Caparrós, P., Llanderal, A., Maksimovic, I., and Lao, M. (2018). Cascade efficiency? J. Clean. Prod. 261:121213. doi: 10.1016/J.JCLEPRO.2020.121213 cropping system with horticultural and ornamental plants under greenhouse Samarakoon, U. C., Fyffe, C., Bale, J., Ling, P., Basnagala, S., Donley, N., et al. conditions. Water 10:125. doi: 10.3390/w10020125 (2019). Effect of electrical conductivity on the productivity and nutrient uptake Grewal, H. S., Maheshwari, B., and Parks, S. E. (2011). Water and nutrient of Lactuca sativa L. Grown using nutrient film technique (NFT). Acta Hortic. use efficiency of a low-cost hydroponic greenhouse for a cucumber crop: an 1266, 137–143. doi: 10.17660/ActaHortic.2019.1266.19 Australian case study. Agric. Water Manag. 98, 841–846. doi: 10.1016/j.agwat. SEDAC (2016). Gridded Population of the World (GPW), v4. Available online at: 2010.12.010 http://sedac.ciesin.columbia.edu/data/collection/gpw-v4 (accessed November Grieve, C. M., and Suarez, D. L. (1997). Purslane (Portulaca oleracea L.): a 27, 2017). halophytic crop for drainage water reuse systems. Plant Soil 192, 277–283. Serio, F., and Elia, A. (2001). Electrical Conductivity of Nutrient Solution. Acta 559, doi: 10.1023/A:1004276804529 563–568. doi: 10.17660/actahortic.2001.559.82 Incrocci, L., Pardossi, A., Malorgio, F., Maggini, R., and Campiotti, C. A. (2003). Shannon, M. C., Grieve, C. M., Lesch, S. M., and Draper, J. H. (2000). Analysis of Cascade cropping system for greenhouse soilless culture. Acta Hortic. 609, salt tolerance in nine leafy vegetables irrigated with saline drainage water. J. Am. 297–301. doi: 10.17660/actahortic.2003.609.44 Soc. Hortic. Sci. 125, 658–664. doi: 10.21273/JASHS.125.5.658 Kruskal, W. H., and Wallis, W. A. (1952). Use of ranks in one-criterion variance Shapiro, S. S., and Wilk, M. B. (1965). An analysis of variance test for normality analysis. J. Am. Stat. Assoc. 47, 583–621. doi: 10.1080/01621459.1952.10483441 (complete samples). Biometrika 52:591. doi: 10.2307/2333709 Lenth, R., Buerkner, P., Herve, M., Love, J., Riebl, H., and Singmann, H. (2020). Singh, H., and Dunn, B. (2016). Electrical Conductivity and pH Guide for Package “emmeans.”. Am. Stat. 34, 216–221. doi: 10.1080/00031305.1980. Hydroponics. Available online at: http://osufacts.okstate.edu (accessed 21 May, 10483031 2020). Levene, H. (1961). Robust Tests for Equality of Variances. Contrib. to Probab. Torrellas, M., Antón, A., López, J. C., Baeza, E. J., Parra, J. P., Muñoz, P., et al. Stat. Essays Honor Harold Hotell. Available online at: https://ci.nii.ac.jp/naid/ (2012). LCA of a tomato crop in a multi-tunnel greenhouse in Almeria. Int. J. 10007628681 (accessed October 23, 2020). Life Cycle Assess. 17, 863–875. doi: 10.1007/s11367-012-0409-8 Magán, J. J., Gallardo, M., Thompson, R. B., and Lorenzo, P. (2008). Effects United Nations (2014). 2014 revision of the World Urbanization of salinity on fruit yield and quality of tomato grown in soil-less culture in Prospects. Available online at: http://www.un.org/en/development/desa/ greenhouses in Mediterranean climatic conditions. Agric. Water Manag. 95, publications/2014-revision-world-urbanization-prospects.html (accessed 1041–1055. doi: 10.1016/j.agwat.2008.03.011 June 20, 2017). Meseguer-Lloret, S. (2017a). Ensayos estadísticos para comparar las pendientes de 2 Weber, C. F. (2016). Nutrient content of cabbage and lettuce microgreens grown on rectas de calibrado: Caso práctico para varianzas homogéneas. Available online vermicompost and hydroponic growing pads. J. Hortic. 03, 1–6. doi: 10.4172/ at: https://riunet.upv.es/handle/10251/84323 (accessed October 13, 2020). 2376-0354.1000190 Meseguer-Lloret, S. (2017b). Ensayos estadísticos para comparar las pendientes de Welch, B. L. (1938). The significance of the difference between two means when the 2 rectas de calibrado: Caso práctico para varianzas NO homogéneas. Available population variances are unequal. Biometrika 29, 350–362. doi: 10.1093/biomet/ online at: https://riunet.upv.es/handle/10251/84444 (accessed October 13, 29.3-4.350 2020). Zhang, C. H., Kang, H.-M., and Kim, I. S. (2006). Effect of using waste nutrient Miceli, A., Moncada, A., and D’Anna, F. (2003). Effect of salt stress in lettuce solution fertigation on the musk melon and cucumber growth. Prot. Hortic. cultivation. Acta Hortic. 609, 371–375. doi: 10.17660/ActaHortic.2003.609.56 Plant Fact. 15, 400–405. Milton, J. S., and Tsokos, J. O. (1983). Statistical Methods in the Biological and Health Sciences. New York, NY: McGraw-Hill. Conflict of Interest: The authors declare that the research was conducted in the Mulder, E. G. (1956). Nitrogen-magnesium relationships in crop plants. Plant Soil absence of any commercial or financial relationships that could be construed as a 7, 341–376. doi: 10.1007/bf01394322 potential conflict of interest. Muñoz, P., Antón, A., Nuñez, M., Paranjpe, A., Ariño, J., Castells, X., et al. (2008). Comparing the environmental impacts of greenhouse versus open-field tomato Copyright © 2020 Rufí-Salís, Parada, Arcas-Pilz, Petit-Boix, Villalba and Gabarrell. production in the mediterranean region. Acta Hortic. 801, 1591–1596. doi: This is an open-access article distributed under the terms of the Creative Commons 10.17660/ActaHortic.2008.801.197 Attribution License (CC BY). The use, distribution or reproduction in other forums Muñoz, P., Flores, J. S. S., Antón, A., and Montero, J. I. I. (2017). Combination is permitted, provided the original author(s) and the copyright owner(s) are credited of greenhouse and open-field crop fertigation can increase sustainability of and that the original publication in this journal is cited, in accordance with accepted horticultural crops in the Mediterranean region. Acta Hortic. 1170, 627–634. academic practice. No use, distribution or reproduction is permitted which does not doi: 10.17660/ActaHortic.2017.1170.78 comply with these terms. Frontiers in Plant Science | www.frontiersin.org 12 November 2020 | Volume 11 | Article 596550