**1. Introduction**

Microgreens can be described as young and tender edible seedlings, produced by using seeds of di fferent vegetable species, herbaceous plants, aromatic herbs, and wild edible plants, which are considered as 'functional foods' or 'super foods' because of their high nutritional value [1–3]. In recent years, microgreens have been increasingly used as basic ingredients in culinary preparations to obtain both sweet and savoury dishes with peculiar organoleptic traits [4]. Many species and local varieties of several botanical families, such as Brassicaceae, can be used for microgreen production [5,6]. The Brassicaceae family o ffers some of the most consumed vegetables worldwide and their seedlings have a generally good taste and high nutritional value. Many studies have been carried out on the nutritional propriety of di fferent Brassicacea genotypes consumed as microgreens.

For example, in a study by Xiao et al. [7], 30 genotypes of *Brassica* were analyzed in regards to the content of elements, while Sun et al. [5] analyzed the polyphenols profile of five Brassicacea species. Other authors [8] also evaluated the bioaccessibility of mineral elements and antioxidant compounds in some Brassicaceae microgreens.

Microgreens can be also used, instead of common vegetables, to reduce the daily intake of some elements when their restriction is required for health reasons. For example, Renna et al. [9] showed that a useful reduction in potassium can occur with three genotypes of microgreens in order to propose low-potassium vegetables for subjects a ffected by renal failure. Recently, many studies were carried out on microgreens in regards to the e ffect of artificial light on carotenoid content [10,11], growth and nutritional quality [12], antioxidant properties [13], and content of bioactive compounds [14]. Nevertheless, only a few studies have been done on the e ffects of nutrient solution strength on the growth and nutritional quality of microgreens [5]. On the other hand, the strength and optimal electric conductibility (EC) of the nutrient solution to maximize yield and content of bioactive compounds, and reduce fertilizer waste during microgreens production, are currently not clear. Some authors [15,16] used a nutrient solution with an EC of 1.12 mS cm<sup>−</sup>1, Kyriacu et al. [17] reported an EC of 0.3 mS cm<sup>−</sup><sup>1</sup> (but with organic substrate with an EC of 0.2 mS cm<sup>−</sup>1), Di Gioia et al. [18] indicated an EC of 1.3 mS cm<sup>−</sup>1, while an EC of 1.8 mS cm<sup>−</sup><sup>1</sup> was reported by Renna et al. [9]. In regards to the chemical composition, some authors [19,20] used a modified Hoagland nutrient solution containing 31.5, 24.2, 6.2, 30.0, 4.1, and 8 mg L−<sup>1</sup> of N, K, P, Ca, Mg, and S, respectively. Di Gioia et al. [21] fertigated microgreens with nutrient solution containing 105.1, 117.4, 15.5, 92.5, 26.0, and 34.6 mg L−<sup>1</sup> of N, K, P, Ca, Mg, and S, respectively, while Wieth et al. [22] used three concentrations (0, 50 and 100%) of a nutrient solution containing 214.2, 250.6, 43.7, 136.0, 26.5, and 35.0 mg L−<sup>1</sup> of N, K, P, Ca, Mg, and S, respectively. The optimal nutrient solution is not clear and much work needs to be done in this area.

An important aspect of the nutritional quality of vegetable products is their nitrate (NO3) content. Nitrate per se is relatively non-toxic, but its reaction products and metabolites, such as nitrite, nitric oxide and N-nitroso compounds have raised concerns because of their implications for adverse health effects, such as methemoglobinemia or 'blue baby syndrome' [23]. In this context, it is interesting to highlight that hydroponic cultivation systems allow a reduction in nitrate content in leafy vegetables, without negatively a ffecting yield and quality, due to strategies such as partially replacing nitrate-based fertilizers with ammonium-based ones [24,25].

Few studies have been carried out until now on the influence of the NH4:NO3 molar ratio in nutrient solutions on mineral and phytochemical content of microgreens. Some authors [17,19,20] reported a NH4:NO3 molar ratio of 11:89 in nutrient solutions, while Wieth et al. [22] used a nutrient solution with a NH4:NO3 molar ratio of 9:91. At the same time, only the NO3 form was used in the nutrient solution by other authors [15,16,21]. Nevertheless, based on the studies carried out on mature vegetables [26–28], it is possible to hypothesize a potential reduction in nitrate content, as well as an improvement in nutraceutical value, in microgreens grown in varying NH4:NO3 molar ratios of the nutrient solution.

Starting from these remarks, the aims of the present study on three *Brassica* microgreens were to evaluate: (i) the e ffects of the nutrient solution strength on yield and quality parameters; and (ii) the physiological behaviour and some quality traits of microgreens fertigated with three di fferent NH4:NO3 molar ratios.

#### **2. Materials and Methods**

#### *2.1. Experimental Set-Up*

Two experiments were conducted using a hydroponic system during the spring of 2015 in the greenhouse at the Experimental Farm 'La Noria' of the Institute of Sciences of Food Production of the Italian National Research Council (CNR), located in Mola di Bari (BA, Southern Italy). The first experiment was carried out from 16 March to 3 April, while the second one was carried out from 22 April to 5 May.

Three different genotypes of Brassicaceae were grown for both experiments: *Brassica rapa* L. subsp. *sylvestris* L. Janch. var. *esculenta* Hort, local variety 'Cima di rapa novantina' (broccoli raab); *Brassica oleracea* L. var. *italica*, cultivar 'Broccolo natalino' (broccoli); *Brassica oleracea* L. var. *botrytis*, cultivar 'Cavolfiore violetto' (cauliflower) (Figure 1). The seeds were purchased from 'Riccardo Larosa Sementi' (Andria, Italy) and their germination, tested at a constant temperature of 20 ◦C, was higher than 95%.

**Figure 1.** Genotypes used for producing microgreens: (**A**) broccoli, cultivar 'Broccolo natalino'; (**B**) broccoli raab, local variety 'Cima di rapa novantina'; (**C**) cauliflower, cultivar 'Cavolfiore violetto'.

Microgreens were grown by using a hydroponic system with polyethylene terephthalate fiber pads (40 cm × 24 cm × 0.89 cm; Sure to Grow®; Sure to Grow, Beachwood, OH, USA) as a growing medium, which was placed on an aluminium bench (180 × 80 cm) with a slope of 0.05%. The seeds were uniformly broadcasted on the surface of the growing media using a seeding density of 4 seeds cm<sup>−</sup>2. The sown fiber pads were irrigated manually using a water-nozzle and were covered with a black polyethylene film until the germination was complete.

During the first experiment, three nutrient solutions (NSs), type-like Hoagland and Arnon [29], with different strengths (1/2 strength, 1/4 strength and 1/8 strength), prepared with rain water were used (Table 1). From germination until harvest, the NS was supplied for one minute in the morning and one minute in the afternoon.


**Table 1.** Characteristics of the nutrient solutions (NS) used during the first experiment.

For the second experiment, three half-strength NS with different ratios of NH4:NO3 were used (Table 2).


**Table 2.** Characteristics of the nutrient solutions used during the second experiment.

To prepare the nutrient solutions, fertilizers for hydroponic production were used. More specifically, the following salts were used: calcium nitrate, potassium nitrate, ammonium nitrate, potassium sulphate, magnesium sulphate, calcium chloride, and potassium dihydrogen phosphate. In order to obtain the element composition reported in Tables 1 and 2, the amount of each salt was calculated, while also considering their titre and purity.

During the second experiment, being in late spring, the temperature in the greenhouse was higher than in the first experiment, for this reason another minute of fertigation was supplied at noon. During both the first and second experiment, after epicotyl emission, NSs were distributed by a drip tape line with pressure-compensated drippers (each with a delivery rate of 0.133 L min−1). An open cycle managemen<sup>t</sup> was used; therefore, the drainage was collected but not reused. The experimental scheme used was split-plot where each plot was represented by the bench and each sub-plot was represented by a genotype.

#### *2.2. Harvesting and Physical Analysis*

Harvesting was carried out by cutting microgreens just above the growing media surface, when the first true leaves were at least 1 cm long. Within each experiment, three samples were considered for each experimental unit (genotype and treatment), and analysed as independent replicates. Each field replicate was obtained by harvesting three sub-samples within the same growing pad.

For both experiments and for each cultivar, we recorded how many days passed from sowing until: breaking seed integuments, radicle spillage, hypocotyl emission, cotyledons formation, first true leaf formation, second true leaf formation (true leaf was formed when it was at least 0.5 cm long). Immediately before the harvesting, other parameters were collected: presence of true leaves, leaf length (true leaf eventually present), shoot height and substrate coverage. To determine presence of true leaves, shoot height and leaf length, three random microgreens were selected for each sub-parcel. The substrate coverage included the distribution and microgreens overlap in the substrate. We used three different categories: 1—low; 2—good; 3—excessive. Each sub-parcel was observed at 30 cm, orthogonally from the growth plan and when possible, between the shoots to watch spare space, where we used category 1. If it was not possible to watch the growth media and there was not any overlap between the shoots, we used category 2, and category 3 was used when there was overlap between the shoots.

The harvested microgreens were weighed to determine the shoot fresh weight (FW) per unit area. The dry matter (DM) was measured in triplicate by oven-drying at 65 ◦C until a constant weight of the samples. The oven-dried samples were used for cation and anion content determination, while freeze-dried (ScanVac CoolSafe 55-9 Pro; LaboGene ApS, Lynge, Denmark) samples were used for chemical analysis.

#### *2.3. Inorganic Ion Content*

The content of inorganic ion was determined by ion exchange chromatography (Dionex DX120; Dionex Corporation, Sunnyvale, CA, USA) with a conductivity detector, as reported by D'Imperio et al. [17]. The content of Na<sup>+</sup>, K<sup>+</sup>, Mg<sup>2</sup>+, and Ca2+ was determined in 1 g of dried sample, using an IonPac CG12A guard column and an IonPac CS12A analytical column (Dionex Corporation); the elution was performed with 18 mM of methanesulfonic acid (Thermo Scientific ™ Dionex ™, Waltham, MA, USA). Peaks identification and calibration were performed using the Multi Element IC Standard solution Fluka TraceCERT ®, Supelco ® (Merck KGaA, Darmstadt, Germany). The contents of Cl− and NO3 − were determined in 0.5 g of dried sample using an IonPac AG14 precolumn and an IonPac AS14 separation column (Dionex Corporation). The eluent consisted of 3.5 mmol·L−<sup>1</sup> of sodium-carbonate (Thermo Scientific ™ Dionex ™, USA) and 1.0 mmol·L−<sup>1</sup> of sodium-bicarbonate solution (Thermo Scientific ™ Dionex ™, USA), and 50 mL of the same eluent was used to extract the anions. Inorganic cation content determination was carried out in triplicate. Peaks identification and calibration were performed using the Multi Element IC Standard sol. IC-MAN-18 (6E) of Chem-Lab (Palin Corporation, Elderslie, UK).

#### *2.4. Dietary Fiber Content*

Dietary fiber content was determined according to AOAC methods [30] with a slight modification. First, a sample of lyophilized microgreen powder (250 mg) was boiled in 32.5 mL of H2SO4 0.64 N for 10 min, adding a few drops of n-octanol as antifoam agent. The resulting insoluble residue was filtered, washed with warm distilled water, and boiled in 32.5 mL of KOH 0.56 N for 10 min. After filtering and washing the sample three times with acetone RPE, it was dried at 105 ± 2 ◦C for 1 h. Weight loss, corresponding to the raw fiber, was determined after cooling the sample at RT in a dryer. Then, ash content was determined by weighing the obtained residue before and after a strong heat treatment (550 ◦C for 3 h). Finally, fiber content was expressed relative to the fresh weight (FW). Crude protein was assessed by the micro-Kjeldahl method, with a nitrogen to protein conversion factor of 6.25, according to the AOAC method 976.05 [30]. Dietary fiber content determination was carried out in triplicate. All chemicals used were supplied by Sigma-Aldrich (Milan, Italy) and were of analytical grade.

#### *2.5. Content of* α*-Tocopherol and* β*-Carotene*

For α-tocopherol and pro-vitamin A expressed as β-carotene, the extraction procedure simultaneously extracts water-soluble vitamin (WSV) and fat-soluble vitamin (FSV). During the extraction process, samples were always protected from direct exposition to light and kept on ice to minimize vitamin degradation. Briefly, 0.050 g of each sample was first extracted with 7.5 mL of 1% BHA in ethanol and 500 μL of internal standard (86.82 μM trans-β-apo-8 carotenal) were added. Samples were placed in an ultrasound bath for 15 s and 180 μL of 80% KOH were added and heated for 45 min at 70 ◦C. Three milliters of water and 3 mL of hexane/toluene were added (10:8 v/v), and centrifuged at 1000× *g* for 5 min. The supernatant was recovered and the bottom solution was extracted with hexane/toluene at least two times. The phases were reunited and the solvent was evaporated under the nitrogen stream. It was recovered with 500 μL of acetonitrile/ethanol 1:1 for HPLC analysis. Separation and identification of lipophilic vitamins in microgreen extracts were carried out with a HPLC 1100 equipped with quaternary pump solvent delivery, thermostatic column compartment, and diode array detector (DAD) (Agilent Technologies, Palo Alto, CA, USA). The samples (20 μL) were injected onto a reversed stationary phase ZORBAX EC18 (Agilent Technologies) (5 μm (150 × 4.6 mm i.d.)), following an isocratic program with ethanol/acetonitrile 1:1 as mobile phase according to the method previously published by Xiao et al. [7]. Stop time was set at 30 min with a re-equilibration time of 10 min corresponding to ~20 column volume (Vc = 0.52 mL). The column temperature was not controlled, while the flow was maintained at 1.2 mL/min. Diode array detection was between 250 nm and 650 nm and absorbance was recorded at 450 nm for β-carotene and 290 nm for α-tocopherol. Compounds identification was achieved by combining di fferent information: positions of absorption maxima (λmax), the degree of vibration fine structure (% III/II), and retention times were compared with those from pure standards. To evaluate linearity, calibration curves with five concentration points for each compound were prepared separately. Calibration was performed by linear regression of peak-area ratios of the vitamins to the internal standard (β-apo-8-carotenal) versus the respective standard concentration, obtaining R<sup>2</sup> values of 0.9992 and 0.9999 for β-carotene and α-tocopherol, respectively. Finally, vitamins were quantified as mg of β-carotene and α-tocopherol per 100 g of microgreens. The determination of α-tocopherol and β-carotene content was carried out in triplicate. All chemicals used were supplied by Sigma-Aldrich (Milan, Italy) and were of analytical grade.

#### *2.6. Statistical Analysis*

The data were analysed by a two-way analysis of variance (ANOVA), using the general linear model procedure of SAS software (SAS Version 9.1, SAS Institute, Cary, NC, USA) and applying a split-plot design with genotype (G) and nutrient solution (NS) as main factors for all measurements. All means were compared using the Student–Newman–Keuls (SNK) test at *p* = 0.05, and standard deviation (SD) was also calculated. Significance of main factors and their interaction are reported in tables. Average values of main factors are reported in tables, while average values of significant interactions G x NS are showed by using histograms.
