Comparative Analysis of Horizontal and Vertical Decoupled Aquaponic Systems for Basil Production and Effect of Light Supplementation by LED

agronomy Article Comparative Analysis of Horizontal and Vertical Decoupled Aquaponic Systems for Basil Production and Effect of Light Supplementation by LED Víctor M. Fernández-Cabanás1,* , Luis Pérez-Urrestarazu 2, Alexes Juárez 3, Nathan T. Kaufman 3and Jackson A. Gross 3 1Urban Greening and Biosystems Engineering Research Group, Departamento de Ciencias Agroforestales, Universidad de Sevilla, ETSIA, Ctra, Utrera km.1, 41013 Seville, Spain 2Urban Greening and Biosystems Engineering Research Group, Area of Agro-Forestry Engineering, Universidad de Sevilla, ETSIA, Ctra, Utrera km.1, 41013 Seville, Spain; lper[email protected] 3Department of Animal Sciences, University of California, Davis, Davis, CA 95616, USA; [email protected] (A.J.); [email protected] (N.T.K.); [email protected] (J.A.G.) *Correspondence: [email protected]; Tel.: +34-954-486-448 Received: 29 July 2020; Accepted: 14 September 2020; Published: 17 September 2020   Abstract: Aquaponic facilities allow the joint production of fish and plants in a reduced space. This hydroponic technique, combined with vertical farming, is not well studied. The main objective of this work was to compare basil production between horizontal and vertical decoupled aquaponic systems and assess the utilization of supplemental lighting in a greenhouse environment. Six independent vertical racks were arranged with hydroponic trays at three heights. Three racks were affixed with LED lighting on the lowest levels and three with supplemental lighting on the intermediate level. Immediately adjacent to the vertical systems, two independent horizontal systems containing three trays were constructed to compare basil production. After 35 days of growth post transplanting, the total production per tray and weight, height, number of leaves, and nodes of 20 selected plants per tray were determined. Records of the intensity of illumination photosynthetic photon flux density (PPFD)) were recorded at dawn (8:00), noon (12:00), and dusk (18:00) at randomly predetermined positions associated with the 20 selected plants per tray. The total basil production in the experiment was 58.79 kg, with an average production per unit area of 2.43 and 0.94 kg m −2 for vertical and horizontal systems, respectively. Productivity per unit area in the vertical systems was 160% greater than in horizontal systems. A significant effect of lighting, the height of the tray, and plant position inside the tray was found on plant growth parameters. Optimization of light source distribution and tray orientation can enhance the productive performance in vertical aquaponic systems. Electricity cost associated with supplemental lighting per kg of production increment was 21.84 and 12.25 $ kg −1 for the bottom and intermediate levels of the vertical system, respectively, the latter being economically the most profitable. Keywords: decoupled aquaponics; vertical farming; hydroponics; basil; artificial lighting 1. Introduction Food production is currently facing significant challenges associated with the limitation of available water and nutrient resources (particularly fertilizers such as phosphorus), changing climatic conditions such as extreme weather, changes in pest distribution, and the increase of high-density urban populations. To address these challenges, approaches such as indoor urban and controlled environmental agriculture have expanded in recent years. One of these emerging modalities is called aquaponics. Aquaponic production combines hydroponic plant production with recirculating Agronomy 2020,10, 1414; doi:10.3390/agronomy10091414 www.mdpi.com/journal/agronomy Agronomy 2020,10, 1414 2 of 14 aquaculture systems to produce both plant and animal protein. Fish and plants grown together use less resources (mainly water and fertilizers) [1,2] in a circular economy framework [3]. Aquaponic facilities can be located on non-arable land within or near urban or peri-urban areas closer to markets, thus shortening the supply chain and reducing the carbon footprint associated with rural farms and transportation of products to city markets [ 4 ]. However, given the land requirements of these facilities, locating them in urban areas is not easily accessible due to the costs of construction and the price of the land. In these situations, vertical farming is an engineering solution to maintain or increase food production using less floor area and extending production into the vertical dimension [ 5 , 6 ]. When comparing vertical and horizontal growing, the following factors must be considered: (1) the management of the system, (2) space requirement, and (3) energy consumption. System management also includes the capital expense and the labor required for the functioning of the system (i.e., planting and harvesting, treatments, maintenance). The space requirement is related to the production costs, available locations, and yield per unit area. The difference in energy consumption is based on the pumping requirements, air movement for carbon dioxide and humidity control, and the need for artificial lighting due to shading caused by a vertical arrangement. Depending on the facility (e.g., greenhouse, indoors) and the vertical system used (e.g., stacked horizontal beds, vertical tower systems, or stepped tiers), the lighting requirements are different. For instance, in stacked horizontal beds, instead of only having one horizontal grow bed, the beds are stacked like shelves in tiers. Hence, only the upper bed would be receiving direct natural light in a greenhouse environment, so supplementary light should be provided for the rest of the levels, usually placed directly under the grow bed above [ 7 ]. As the need for artificial lighting in vertical production leads to an increase in energy costs, a return on investment must lead to a higher yield per unit floor area. In this sense, the development of LED horticultural lights, long-lasting and energy-efficient, has increased the competitiveness of this type of solution [4]. Few studies are assessing the difference between horizontal and vertical hydroponic vegetable production [ 5 , 8 , 9 ], and none are comparing vertical vs. horizontal aquaponic systems [ 7 ]. Therefore, the main objective of this work was the comparison between horizontal and vertical production in terms of total yield and quality of basil using a decoupled aquaponic system. In this growing modality, the hydroponic system used recirculating aquaculture wastewater similarly as traditional hydroponic nutrients are applied; the distinction was that the water did not return to the fish from the plant production system. As a secondary objective, we aimed to assess the effect of additional artificial lighting in the vertical systems on total production, plant development, and energy consumption. 2. Materials and Methods 2.1. System Description and Experimental Design In this investigation, six decoupled aquaponic vertical racks were compared (Figure 1) to two horizontal systems (Figure 2). Each vertical rack was composed of three 1.22 × 2.44 × 0.15 m flood and drain trays (Botanicare, Chandler, AZ, USA), a 473 L sump filled with effluents coming from three 2000 L recirculating fish tanks, and a 4164 L h −1 submersible water pump (ActiveAqua 1100 GPH, Hydrofarm, Petaluma, CA, USA). Each tray was irrigated individually (i.e., did not drain from one to the next), and flow was regulated with ball valves to have equal flow to fill all trays. These vertical rack systems were placed roughly 1.22 m apart and contained three shelves, with one of the two lower shelves receiving LED lighting distributions. Trays were spaced 0.61 m apart vertically, with shelving at the heights of 0.61 m (Level 1), 1.22 m (Level 2), and 1.83 m (Level 3). Trays were flood-irrigated twice a day, at 07:00 and 14:00, for 15 min using a timer connected to pumps, to a depth of approximately 0.02 m. All trays contained 200 individual plants, with a single plant per pot, forming a 20 ×10 grid. Agronomy 2020,10, 1414 3 of 14 Agronomy 2020, 10, x FOR PEER REVIEW 3 of 14 Figure 1. Schematics of the vertical racks with a random distribution of light supplementation in level 1 (0.61 m height) and 2 (1.22 m height). The plant placement is not an accurate reflection of individual plants in the system. Figure 2. Schematics of (a) elevation and (b) plant of the horizontal systems. The plant placement is not an accurate reflection of individual plants in the tray. All top shelves were exposed to natural light, with the lower shelves with or without light supplementation by four Infinity 2.0 (Thrive Agritech, New York, NY, USA) LED light bars (1.2 m length, 65 W, 145 µmol s−1 light output) or had only natural light. LED bars provided full-spectrum white light with peaks at wavelengths corresponding to blue and red (Figure 3). This assay design blocked for greenhouse location, shelf height, and lighting. The prescribed artificial light in the vertical systems was randomized by number generator to alternate one LED-supplemented and one sunlight shelf per rack at levels 1 and 2. Supplemental lighting operated 14 h per day, from 06:30 to 20:30. Figure 3. Spectral features (photosynthetic photon flux (PPF) values) of Infinity 2.0 LED bars used in the study (https://www.thriveagritech.com/wp-content/uploads/2020/05/Infinity-2.0-Datasheet.pdf). (a) (b) Figure 1. Schematics of the vertical racks with a random distribution of light supplementation in level 1 (0.61 m height) and 2 (1.22 m height). The plant placement is not an accurate reflection of individual plants in the system. Agronomy 2020, 10, x FOR PEER REVIEW 3 of 14 Figure 1. Schematics of the vertical racks with a random distribution of light supplementation in level 1 (0.61 m height) and 2 (1.22 m height). The plant placement is not an accurate reflection of individual plants in the system. Figure 2. Schematics of (a) elevation and (b) plant of the horizontal systems. The plant placement is not an accurate reflection of individual plants in the tray. All top shelves were exposed to natural light, with the lower shelves with or without light supplementation by four Infinity 2.0 (Thrive Agritech, New York, NY, USA) LED light bars (1.2 m length, 65 W, 145 µmol s−1 light output) or had only natural light. LED bars provided full-spectrum white light with peaks at wavelengths corresponding to blue and red (Figure 3). This assay design blocked for greenhouse location, shelf height, and lighting. The prescribed artificial light in the vertical systems was randomized by number generator to alternate one LED-supplemented and one sunlight shelf per rack at levels 1 and 2. Supplemental lighting operated 14 h per day, from 06:30 to 20:30. Figure 3. Spectral features (photosynthetic photon flux (PPF) values) of Infinity 2.0 LED bars used in the study (https://www.thriveagritech.com/wp-content/uploads/2020/05/Infinity-2.0-Datasheet.pdf). (a) (b) Figure 2. Schematics of ( a ) elevation and ( b ) plant of the horizontal systems. The plant placement is not an accurate reflection of individual plants in the tray. All top shelves were exposed to natural light, with the lower shelves with or without light supplementation by four Infinity 2.0 (Thrive Agritech, New York, NY, USA) LED light bars (1.2 m length, 65 W, 145 µ mol s −1 light output) or had only natural light. LED bars provided full-spectrum white light with peaks at wavelengths corresponding to blue and red (Figure 3). This assay design blocked for greenhouse location, shelf height, and lighting. The prescribed artificial light in the vertical systems was randomized by number generator to alternate one LED-supplemented and one sunlight shelf per rack at levels 1 and 2. Supplemental lighting operated 14 h per day, from 06:30 to 20:30. Agronomy 2020, 10, x FOR PEER REVIEW 3 of 14 Figure 1. Schematics of the vertical racks with a random distribution of light supplementation in level 1 (0.61 m height) and 2 (1.22 m height). The plant placement is not an accurate reflection of individual plants in the system. Figure 2. Schematics of (a) elevation and (b) plant of the horizontal systems. The plant placement is not an accurate reflection of individual plants in the tray. All top shelves were exposed to natural light, with the lower shelves with or without light supplementation by four Infinity 2.0 (Thrive Agritech, New York, NY, USA) LED light bars (1.2 m length, 65 W, 145 µmol s−1 light output) or had only natural light. LED bars provided full-spectrum white light with peaks at wavelengths corresponding to blue and red (Figure 3). This assay design blocked for greenhouse location, shelf height, and lighting. The prescribed artificial light in the vertical systems was randomized by number generator to alternate one LED-supplemented and one sunlight shelf per rack at levels 1 and 2. Supplemental lighting operated 14 h per day, from 06:30 to 20:30. Figure 3. Spectral features (photosynthetic photon flux (PPF) values) of Infinity 2.0 LED bars used in the study (https://www.thriveagritech.com/wp-content/uploads/2020/05/Infinity-2.0-Datasheet.pdf). (a) (b) Figure 3. Spectral features (photosynthetic photon flux (PPF) values) of Infinity 2.0 LED bars used in the study (https://www.thriveagritech.com/wp-content/uploads/2020/05/Infinity-2.0-Datasheet.pdf). Agronomy 2020,10, 1414 4 of 14 The two horizontal systems, located immediately east of vertical racks (Figure 4), consisted of the same equipment as vertical systems with identical trays (3 trays per rack), sumps, pumps, and irrigation periods. Plants in horizontal trays received natural sunlight only Agronomy 2020, 10, x FOR PEER REVIEW 4 of 14 The two horizontal systems, located immediately east of vertical racks (Figure 4), consisted of the same equipment as vertical systems with identical trays (3 trays per rack), sumps, pumps, and irrigation periods. Plants in horizontal trays received natural sunlight only The experiment was carried out from 19 August to 8 October 2019 in a 280 m2 polycarbonate greenhouse located at UC Davis Campus (University of California, Davis, CA, USA; 38°32′18.8″ N 121°46′54.2″ W). Figure 4. Distribution of vertical and horizontal systems inside the greenhouse. 2.2. Fish and Plant Species On the campus of the University of California, Davis (Davis, CA, USA), one of the products of interest is basil (Ocimum basilicum), widely used to make pesto sauce in different dishes on the daily menu for the Dining Commons. It is also a popular specialty herb that does well in commercial greenhouse production systems. For these reasons, basil was chosen as the plant species for this trial. However, since the available horizontal space limits the production of large commercial scale, vertical production was selected to compare the vertical growing systems’ effectiveness with that of conventional horizontal systems. Basil is commonly produced using a multitude of commercial hydroponic growing methods that utilize an assortment of irrigation techniques. Commercial methods include the nutrient film technique (NFT), deep water culture (DWC), and, as was the case with this study, flood and drain methodology with pots filled with a soilless growing medium. Vegetable propagation trays (338S67S, Proptek, Watsonville, CA, USA) were filled with stabilized rooting media (Q plugs®, International Horticulture Technologies, Hollister, CA, USA) and hydrated with reverse osmosis (RO) water. Pelleted seeds of Nufar Fusarium-resistant Italian large-leaf type basil (Johnny’s Selected Seeds, Inc., Winslow, ME, USA) were individually sown into each cell and placed into a flood and drain propagation rack. During germination, the propagation trays were irrigated with RO water twice a day for one hour (06:00–07:00 and 14:00–15:00). Fourteen days after germination, 4800 individual seedlings were transplanted into 0.102 × 0.102 × 0.089 m3 pots with ground Q plugs media and then Figure 4. Distribution of vertical and horizontal systems inside the greenhouse. The experiment was carried out from 19 August to 8 October 2019 in a 280 m 2 polycarbonate greenhouse located at UC Davis Campus (University of California, Davis, CA, USA; 38 ◦ 32 0 18.8 00 N 121◦46054.200 W). 2.2. Fish and Plant Species On the campus of the University of California, Davis (Davis, CA, USA), one of the products of interest is basil (Ocimum basilicum), widely used to make pesto sauce in different dishes on the daily menu for the Dining Commons. It is also a popular specialty herb that does well in commercial greenhouse production systems. For these reasons, basil was chosen as the plant species for this trial. However, since the available horizontal space limits the production of large commercial scale, vertical production was selected to compare the vertical growing systems’ effectiveness with that of conventional horizontal systems. Basil is commonly produced using a multitude of commercial hydroponic growing methods that utilize an assortment of irrigation techniques. Commercial methods include the nutrient film technique (NFT), deep water culture (DWC), and, as was the case with this study, flood and drain methodology with pots filled with a soilless growing medium. Vegetable propagation trays (338S67S, Proptek, Watsonville, CA, USA) were filled with stabilized rooting media (Q plugs ® , International Horticulture Technologies, Hollister, CA, USA) and hydrated with reverse osmosis (RO) water. Pelleted seeds of Nufar Fusarium-resistant Italian large-leaf type basil (Johnny’s Selected Seeds, Inc., Winslow, ME, USA) Agronomy 2020,10, 1414 5 of 14 were individually sown into each cell and placed into a flood and drain propagation rack. During germination, the propagation trays were irrigated with RO water twice a day for one hour (06:00–07:00 and 14:00–15:00). Fourteen days after germination, 4800 individual seedlings were transplanted into 0.102 × 0.102 × 0.089 m 3 pots with ground Q plugs media and then placed into their respective systems at a density of 200 containers per tray (around 67 plants m −2 ). Each vertical rack system was supplemented with an additional 110 L of sturgeon effluent twice a week to replenish losses from evapotranspiration and maintain nutrient levels in the water. Biological nutrients were sourced from the three recirculating aquaculture systems (RASs) located at UC Davis Center for Aquatic Biology and Aquaculture (CABA). White (Acipenser transmontanus) and green (Acipenser medirostris) sturgeon (<1 yr) were reared in 2000 L fish tanks on well water with mechanical and biological filtration occurring by bead filters. Tanks were fed 200 g daily of sinking complete feed for trout and steelhead (Skretting, Tooele, UT, USA) with 46% of crude protein, 12% of crude fat, 3% of crude fiber, and 1.1% of phosphorous. Water changes occurred twice a week by pumping 2200 L of filtered aquaculture tank water into a storage tank and transporting the water to the greenhouse. 2.3. System Operations and Measurements Performed 2.3.1. Plant Nutrient Supplementation Aquaponic systems usually show low concentrations of K, Fe, P, or Ca in the water soluble fraction of fish effluents [ 10 , 11 ]. Therefore, the plants were visually examined to observe for the occurrence of nutrient insufficiency, and seedlings were prescribed a 1% KNO 3 foliar feed treatment. Plants were dosed in the morning twice a week with a manual sprayer with additional potassium without affecting the pH of the nutrient solution. Iron is also one of the essential micronutrients usually deficient in aquaponic systems due to its low amount in commercial fish feed [ 12 ]. Therefore, 0.2 g of chelated iron diethylene triamine pentaacetic acid (DTPA) (13% Fe) was dissolved weekly in the water of each sump, resulting in an increase in the Fe concentration of 0.055 mg L −1 . Thus, the total Fe applied to each system was 0.182 g, while the expectable Fe uptake by plants for optimum nutrition was estimated to be around 0.174 g. Fe–DTPA is assumed to remain in solution available to plants at the pH range of the solution in sumps. This calculation allows us to suppose that the Fe supply was enough to cover plant needs without increasing Fe concentration in sumps to toxic levels, an argument supported by the total absence of symptoms of internerval chlorosis. In this sense, crop needs were covered with increases in Fe concentration in sumps lower than that previously reported for basil production in the University of the Virgin Islands (UVI) system (2 mg L −1 DTPA applied every 3 weeks) [ 13 ]. Hydroponic plant production is traditionally grown in a soilless medium with the majority of nutrients provided by an aqueous solution. In this study, all nutrients were provided by flood irrigation, supplied directly to saturate the root zone with the exception of potassium, which was applied via foliar spray. 2.3.2. Water Quality Nitrate concentrations (ppm) were quantified weekly using a LaMotte ® Nitrate Nitrogen tablet testing kit (Chestertown, MD, USA) to maintain a concentration around 60 ppm (57.5 ± 18.4 ppm) in all systems after fish effluent replenishment. Weekly nitrate uptake in each sump varied across treatments and replicates. When low values were recorded, sumps were replenished with fish effluents to increase levels to the desired range, which in aquaponic systems is lower than that required for hydroponic systems [ 14 ]. Daily water temperature, pH, and electrical conductivity (EC) readings were recorded using a Bluelab ® combo meter (Bluelab Corporation Limited, Tauranga, New Zealand). A combination of phosphoric and citric acid (pH Down; General Hydroponics, Sebastopol, CA, USA) lowered pH to the desired 6.0–6.5 range. Agronomy 2020,10, 1414 6 of 14 2.3.3. Lighting Measurements Photosynthetic photon flux density (PPFD) measurements were taken biweekly in the morning (8:00), afternoon (12:00), and evening (18:00), for the 20 selected pots in each tray, using a full-spectrum quantum meter (MQ-500; Apogee Instrument, Logan, UT, USA). Readings were performed at the current height of the plants to assess light exposure on the plants throughout the day and to quantify the difference between natural and artificial light sources at the different levels of shelves. Average light readings at each position were correlated with plant yield (weight, height, number of leaves, and number of nodes). Data collection at all plant locations (discussed in the next section) required, on average, 1.75 h. 2.3.4. Plant Yield For each tray, 20 individual plant sites were monitored during plant development throughout the five-week study and examined biweekly for height, number of leaves, and number of nodes. Plant selection (Figure 5) was determined by dividing the tray of two hundred plants into two squares of 10 × 10 containers to account for potential variability of temperature with the proximity to the wet wall and of lighting. In each square, for consecutive positions in the north–south direction, a random value between 1 and 10 was generated. If the number was already selected, a new number was generated. Each square was treated independently so that only one pot was selected per line in the east–west direction, whereas two pots were selected in each line in a north–south direction. The same plant positions were selected for each of the 24 tray locations in this study. All basil, 35 days after transplanting (when lighting height was reached), was harvested at the base of the plant above the ground portion. The total biomass per tray was weighed, and 20 randomly predetermined plants per tray were weighed, measured, and the number of leaves and nodes quantified. Agronomy 2020, 10, x FOR PEER REVIEW 6 of 14 Average light readings at each position were correlated with plant yield (weight, height, number of leaves, and number of nodes). Data collection at all plant locations (discussed in the next section) required, on average, 1.75 h. 2.3.4. Plant Yield For each tray, 20 individual plant sites were monitored during plant development throughout the five-week study and examined biweekly for height, number of leaves, and number of nodes. Plant selection (Figure 5) was determined by dividing the tray of two hundred plants into two squares of 10 × 10 containers to account for potential variability of temperature with the proximity to the wet wall and of lighting. In each square, for consecutive positions in the north–south direction, a random value between 1 and 10 was generated. If the number was already selected, a new number was generated. Each square was treated independently so that only one pot was selected per line in the east–west direction, whereas two pots were selected in each line in a north–south direction. The same plant positions were selected for each of the 24 tray locations in this study. All basil, 35 days after transplanting (when lighting height was reached), was harvested at the base of the plant above the ground portion. The total biomass per tray was weighed, and 20 randomly predetermined plants per tray were weighed, measured, and the number of leaves and nodes quantified. In order to prevent propagation error and to minimize experimental bias, a number of methods were conducted during the essay: (1) daily plant growth measurements and data recording were conducted by the same individual throughout the experiments; (2) final measurements and data recording were conducted by the same individual throughout the data collection experiments; (3) all data were recorded blind to LED treatments as final data collection was conducted in a separate room outside the view of the plant harvest; and (4) during data collection, the data recorder repeated and verified each record during the transcribing of data into a notebook, and all data recorded in the electronic spreadsheets were verified against the written copy to prevent input errors. Figure 5. Distribution of 20 selected containers in each tray. 2.4. Statistical Analysis Statistical analyses of the data were performed with the STATGRAPHICS Centurion XVIII statistical package (version 18.1.12, StatPoint Technologies Inc., Warrenton, VA, USA). For statistical comparison, a generalized linear model (GLM) was used. The normality of the data was assessed by means of the Shapiro–Wilk test. When data were not normally distributed, a nonparametric ANOVA (Kruskal–Wallis test) was used to confirm significant differences. Post hoc analysis was performed using the Honestly-significant-difference (HSD) Tukey test and the Games–Howell test (when variances were not equal). Differences were considered significant when p < 0.05. Figure 5. Distribution of 20 selected containers in each tray. In order to prevent propagation error and to minimize experimental bias, a number of methods were conducted during the essay: (1) daily plant growth measurements and data recording were conducted by the same individual throughout the experiments; (2) final measurements and data recording were conducted by the same individual throughout the data collection experiments; (3) all data were recorded blind to LED treatments as final data collection was conducted in a separate room outside the view of the plant harvest; and (4) during data collection, the data recorder repeated and verified each record during the transcribing of data into a notebook, and all data recorded in the electronic spreadsheets were verified against the written copy to prevent input errors. 2.4. Statistical Analysis Statistical analyses of the data were performed with the STATGRAPHICS Centurion XVIII statistical package (version 18.1.12, StatPoint Technologies Inc., Warrenton, VA, USA). For statistical Agronomy 2020,10, 1414 7 of 14 comparison, a generalized linear model (GLM) was used. The normality of the data was assessed by means of the Shapiro–Wilk test. When data were not normally distributed, a nonparametric ANOVA (Kruskal–Wallis test) was used to confirm significant differences. Post hoc analysis was performed using the Honestly-significant-difference (HSD) Tukey test and the Games–Howell test (when variances were not equal). Differences were considered significant when p<0.05. 3. Results Average basil production recorded per unit area was 2.43 and 0.94 kg m −2 for vertical and horizontal systems, respectively. Production per unit area in each of the trays was assessed to evaluate the differences between trays in different levels and lighting. Basil production in trays located at tray level 1 showed significant differences according to lighting, with the highest values corresponding to shelves supplemented with LED, while no significant differences were found at level 2 (Table 1). However, no significant differences were found between levels 1 and 2 for the same lighting conditions, although higher average values were obtained at level 2. Tray level 3 showed higher productivity than level 1, with only natural sunlight. Furthermore, comparing the production values obtained for levels 1 and 2 and taking into account only the height position (tray level) or the lighting mode (Table 2), we found that both factors had a significant influence, while their interaction did not. Table 1. Mean values of basil production (kg m −2 ) in vertical racks according to tray level and lighting. Means with the same lowercase letter in each column and with the same capital letter in each row are not significantly different at the p<0.05 level, according to the HSD Tukey test. Basil Production (kg m−2) Level 1 Level 2 Level 3 p Sunlight 0.451 aA 0.685 aAB 0.981 B 0.0092 Sunlight +LED 0.750 bA 1.191 aA 0.0798 p0.002 0.1106 Table 2. Mean values ( ± SE) of basil production (kg m −2 ) in levels 1 and 2 of vertical systems according to height and lighting. Means with the same letter in each column are not significantly different at the p<0.05 level, according to the HSD Tukey test. Lighting p=0.0070 Production (kg m−2) Height p=0.0229 Production (kg m−2) Sunlight 0.568 ±0.087 a Level 1 0.601 ±0.087 a Sunlight +LED 0.971 ±0.087 b Level 2 0.938 ±0.087 b Production in the systems exposed to natural lighting found no significant differences (p=0.6404) between the horizontal systems (0.939 ± 0.058 kg m −2 ) and the production obtained in the upper level of the vertical systems (level 3) (0.981 ± 0.064 kg m −2 ). When comparing production per unit area for each of the two types of systems studied (vertical and horizontal), significant differences were found in favor of vertical systems, as expected (Table 3). Table 3. Mean values ( ± SE) of basil production (kg m −2 ) in horizontal systems and vertical racks. Means with the same letter in each column are not significantly different at the p<0.05 level, according to the HSD Tukey test. Basil Production (kg m−2) Horizontal 0.939 ±0.246 a Vertical 2.427 ±0.155 b p=0.0037 Agronomy 2020,10, 1414 8 of 14 To more thoroughly evaluate factors of plant development in the horizontal and vertical systems, a GLM analysis was performed to determine the effect of the tray level, the lighting mode, and the position of the plant within the trays, using the data (weight, height, number of leaves, and number of nodes) obtained from the 20 selected basil plant individuals at harvest. Similar to the analysis of productivity, a first study was carried out for tray levels 1 and 2 of the vertical systems where the level (1 and 2) and the type of lighting (natural or with LED supplement) were considered as categorical factors, while the distance to the north end of the tray (north–south direction) and the distance to the east end (east–west direction) were treated as quantitative factors. Out of quantitative factors, only the north–south direction showed a significant effect for all parameters, whereas both categorical factors were significant. The interaction between level and lighting system was also significant for all the parameters (Table 4), except for the number of nodes (Table 5). Higher values for all parameters are associated with the upper level in trays and with LED supplementation, including now a positive effect for plants located at southern positions in trays. All results obtained for individual plant parameters seem to be consistent with those from productivity in trays. Nevertheless, due to the higher number of replicates and to the information related to the plants’ position in the trays, significant effects were found for categorical factors, their interaction, and the position of plants in the north–south direction. Table 4. Mean values ( ± SE) of weight, height, and number of leaves of 20 selected plants in vertical racks at levels 1 and 2. Means with the same letter in each column are not significantly different at the p<0.05 level, according to the HSD Tukey test. Plant Weight (g) Plant Height (cm) Number of Leaves Level 1 ×Sunlight 7.13 ±0.46 a 20.04 ±0.55 a 9.30 ±0.62 a Level 2 ×Sunlight 8.53 ±0.46 a 26.13 ±0.55 b 11.28 ±0.62 a Level 1 ×LED 11.52 ±0.46 b 20.48 ±0.55 a 16.82 ±0.62 b Level 2 ×LED 17.04 ±0.46 c 33.23 ±0.55 c 24.65 ±0.62 c Level ×lighting p=0.0000 p=0.0000 p=0.0000 North-South direction p=0.0000 p=0.0000 p=0.0000 Table 5. Mean values ( ± SE) of the number of nodes of 20 selected plants in vertical racks at levels 1 and 2. Means with the same letter in each column are not significantly different at the p<0.05 level, according to the HSD Tukey test. Number of Nodes Height p=0.0000 Level 1 4.49 ±0.06 a Level 2 5.22 ±0.06 b Lighting p=0.0000 Sunlight 4.55 ±0.06 a Sunlight +LED 5.16 ±0.06 b North-South direction p=0.0000 The upper tray of vertical racks was again compared with horizontal trays using selected plant parameters at harvest. Even though no differences were obtained previously in the comparison of the productivities of the trays when the weights, heights, and number of nodes of the selected plants were analyzed, significant differences were found in favor of the vertical systems. There was also a significant effect of the north–south position on weight and height (Table 6). Only the number of leaves showed no significant differences in both systems. Agronomy 2020,10, 1414 9 of 14 Table 6. Mean values ( ± SE) of weight, height, number of leaves, and number of nodes of 20 selected plants in trays of horizontal systems and level 3 (upper level) of vertical racks. Means with the same letter in each column are not significantly different at the p<0.05 level, according to the HSD Tukey test. Plant Weight (g) Plant Height (cm) No. of Leaves No. of Nodes Horizontal 12.98 ±0.30 a 31.58 ±0.45 a 22.73 ±0.50 a 5.70 ±0.05 a Vertical 13.97 ±0.33 b 33.55 ±0.49 b 24.06 ±0.55 a 6.18 ±0.05 b System p=0.0293 p=0.0033 p=0.0745 p=0.0000 North-South direction p=0.0177 p=0.0468 p>0.05 p>0.05 Since differences in the intensity of the illumination affect plant growth and yield parameters, the mean values and standard deviations of PPFD records were calculated for the 20 positions of each tray. Table 7shows the values measured at dawn (8:00), noon (12:00), and dusk (18:00) with the average values from trays in horizontal and vertical systems separated by quadrant, namely north (samples 1 to 7), center (8 to 13), and south (14 to 20). Higher PPFD values are observed for the three selected times at higher levels, LED-supplemented trays, and southern positions in trays. Table 7. PPFD mean and SD values ( µ mol s −1 m −2 ) were recorded at dawn, noon, and dusk at 20 selected positions per tray. Mean values were calculated by tray quadrant, namely north (samples 1 to 7), center (8 to 13), and south (14 to 20). V1, V2, and V3: Vertical trays at levels 1, 2, and 3, respectively. H: horizontal tray. SUN: only natural sunlight. LED: with LED supplementation. Dawn Noon Dusk Average Tray Light Position Mean SD Mean SD Mean SD Mean SD V1 SUN North 32.9 13.9 77.4 31.3 24.1 8.5 44.8 30.9 V1 SUN Center 31.1 15.0 72.1 23.4 21.4 6.8 41.5 27.5 V1 SUN South 44.7 18.7 183.3 110.1 43.3 42.7 90.4 95.1 V1 LED North 170.5 40.8 208.2 29.7 142.6 39.2 173.8 45.5 V1 LED Center 168.4 33.5 204.8 27.4 145.1 37.6 172.8 41.0 V1 LED South 186.3 52.3 290.1 91.8 141.9 45.5 206.1 90.7 V2 SUN North 57.9 16.1 101.0 26.8 19.1 14.9 59.3 39.0 V2 SUN Center 57.8 16.2 97.4 19.6 17.1 9.8 57.4 36.4 V2 SUN South 79.9 21.6 193.5 80.4 36.3 33.9 103.3 84.1 V2 LED North 168.9 47.7 224.6 33.3 167.2 66.1 186.9 57.1 V2 LED Center 165.0 34.7 216.2 25.9 161.9 48.5 181.0 44.8 V2 LED South 170.9 48.1 280.6 67.2 164.3 58.3 205.2 78.9 V3 SUN North 130.9 25.5 361.9 31.1 98.9 81.3 197.2 128.5 V3 SUN Center 143.5 34.3 416.9 67.1 127.8 94.7 229.4 150.1 V3 SUN South 154.2 38.5 467.8 49.4 117.4 80.7 246.5 168.2 H SUN North 408.1 274.2 399.0 87.5 23.2 3.8 276.8 244.3 H SUN Center 446.2 282.0 406.9 69.0 24.6 5.4 292.6 253.3 H SUN South 438.7 283.5 425.2 84.8 25.7 6.3 296.5 256.5 To evaluate the influence of illumination on yield, the average value of all PPFD readings were regressed with plant weight using a multiplicative model, and a significant correlation between both parameters (p=0.0000) was found with a coefficient of determination of 0.52. The equation corresponding to the adjusted model is as follows: ln(Weight) =−0.422383 +0.571854 ×ln(PPFD). (1) Adjustment with a multiplicative regression model was made as the recorded PPFD data had average values well below the 1600 µmol s−1m−2when saturation was reached. To study the economic feasibility of lighting supply, the productivity of racks with supplemental lighting in tray level 1 (LED-1) and level 2 (LED-2) was compared with the estimated productivity of a