*2.1. Experimental Site, Plant Material, and Management Practices*

The experiment was conducted in 2014, in a greenhouse situated in the coastal area of Eastern Sicily (37◦24 26" N, 15◦03 37" E, 6 m a.s.l.). The local climate is semi-arid/Mediterranean, with mild and wet winters, and hot, dry summers. A 1000 m2, east–west-oriented, multi-aisle greenhouse was used, having a steel tubular structure and covered with polycarbonate slabs. Mean air temperature, relative humidity and global radiation inside the greenhouse (two sets of sensors in the center of each experimental plot) were recorded on a data logger (CR10 X; Campbell Scientific Ltd., Loughborough, UK). Melon cv. "Proteo" F1 (Syngenta Seed, Basel, Switzerland) belonging to the Reticulatus group was used as scion. Five F1 rootstock genotypes were included in the study, of which 2 were intraspecific, namely "Dinero" (Syngenta Seed) and "Magnus" (Agris), and 3 were interspecific (*Cucurbita maxima* Duchesne × *C. moschata* Duchesne ex Poir.), namely "RS841" (DeRuiter Seeds, Oxnard, CA, USA) "Shintoza" (Fenix, Belpasso, Italy) and 'Strong Tosa' (Syngenta Seed). Self-grafted "Proteo" was used as control. Splice-grafted plantlets were used, whereas plastic clips were applied to secure the creation of the graft union. Plantlets were obtained from a specialized nursery and transplanted at the stage of 3 true leaves on April 22, in 5 L pots filled with perlite (3–5 mm). Pots were placed in troughs (5 per main plot) of an open soilless system, placed at 5 cm from the soil surface and with a distance of 50 cm between pots and 100 cm between troughs, obtaining a plant density of 2 plants m<sup>−</sup>2. During the trial, the crop was fertigated with a nutrient solution having the following composition, including the starting well water (mmol L−1): 11.2 NO3 <sup>−</sup>, 0.3 NH4 <sup>+</sup>, 1.3 H2PO4 <sup>−</sup>, 6.6 K<sup>+</sup>, 0.9 SO4 <sup>2</sup>−, 3.4 Ca2<sup>+</sup>, 2.5 Mg2<sup>+</sup>. The concentration of microelements (μmol L−1) was: 15 Fe3<sup>+</sup>, 10 Mn2+, 0.75 Cu2+, 5 Zn2+, 30 B3<sup>+</sup>, 0.5 Mo6<sup>+</sup>. The pH was maintained at 5.9 by adding H2SO4 (95% concentration, 1.83 kg L−1). Bumblebees were introduced into the greenhouse to maximize pollination. Stems were left to grow horizontally, whereas plants were pruned by removing the lateral shoots. Only 1 lateral ramification

per plant was left, bearing 2 fruits. In order to evaluate the effects of studied factors at the same date, the trial was stopped when the first fruit was fully ripe (i.e., 45 days after transplant). Fifteen days after transplanting, the nutrient solution was differentiated to obtain 2 As concentrations, a control solution (0.002 mg L−<sup>1</sup> As, which was the concentration in the starting fertigation solution, hereafter As−) and an As-enriched one (3.8 mg L−<sup>1</sup> As, hereafter As+) obtained by adding sodium arsenate (Na2HAsO4 7H2O, 24% As content). The As+ treatment was chosen to simulate the working condition of a soil having 55 ppm of As, ~7.0% of which was bioavailable (a common situation in As contaminated soils). The nutrient solutions were supplied using a drip irrigation system with one emitter per plant (4 L h<sup>−</sup>1). The amount of nutrient solution supplied at each irrigation was quantified on a weekly basis, according to the volume of the substrate exploited by the roots and the corresponding water contained in the substrate at the intervals from <sup>−</sup>10 to <sup>−</sup>50 hPa of matrix potential (12 mL 100 mL<sup>−</sup>1).

The experiment was arranged in a randomized split-plot design with three replicates, assigning the As concentration of the nutrient solution to the main plots, and the rootstock combination to the subplots. The overall experimental area inside the greenhouse was 450 m2 (15.0 <sup>×</sup> 30.0 m), including 900 plants (324, excluding border plants), divided into 36 net experimental units (2 As levels × 6 grafting combinations × 3 replicates) each containing 9 plants.

#### *2.2. Leaf Relative Chlorophyll (Chl) Content and Gas Exchange Measurements*

Forty-five days after transplanting plants were checked for leaf relative Chl content, through a portable Chl meter (SPAD 502; Minolta Camera, Osaka, Japan). Before measurements, the instrument was calibrated according to manufacturer's instructions. All readings were taken from the adaxial side of the tallest fully expanded leaf. To minimize possible interactions with either plant water status and natural irradiance level [21,22], measurements were made in the morning, starting at 08:00 h (local solar time). Instantaneous leaf photosynthetic rate (AN), stomatal conductance (gs), and leaf transpiration rate (E) were also measured from 11:00 to 13:00 inside a 6.25 cm<sup>2</sup> leaf chamber of a portable photosynthesis system (LCi; ADC BioScientific Ltd., Hoddesdon, UK). Photosynthetic water use efficiency (WUE) was calculated as the ratio AN/E [23]. During measurements, leaf temperature was 27.4 ± 2 ◦C, while mean photosynthetic photon flux density was in the range of 1500 ± 100 μmol photons m−<sup>2</sup> s<sup>−</sup>1. Duplicate measurements were taken from four plants per sub-plot.

#### *2.3. Plant Growth and Development Measurements*

On the same date of physiological measurements, all the plants within each replicate were harvested and their main fractions (leaves, stem and fruits, regardless of the ripening stage) weighed separately. The number of leaves per plant (LN) was determined, while plant leaf area (LA) was measured using an Image Analysis System (DeltaT Devices Ltd., Cambridge, UK), then subsamples of raw materials were kept in a thermo-ventilated oven at 70 ◦C (Binder, Milan, Italy) until constant weight, in order to determine their dry weight (DW). From the original data frame, the leaf area ratio (LAR, the ratio between the area and total plant biomass) and leaf weight ratio (LWR, the dry weight of leaves to whole plant dry weight) were calculated.

### *2.4. Arsenic and Phosphorous Determination in Plant Tissues*

Roots, shoots, and fully ripe fruits (1 per plant) were analyzed to determine the quantity of the total As and phosphorus (P). To this end, about 200 mg of samples were subjected to acid digestions and to As and P determinations, which were performed according to Stazi et al. [19]. The reagents were super pure for trace analysis. The accuracy of the measurements was assessed using SRM 1573a as standard reference materials trace metals. Total As quantification were performed using an inductively-coupled plasma optical emission spectrometer (ICP-OES) with an axially viewed configuration (8000 DV, PerkinElmer, Shelton, CT, USA) equipped with an ultrasonic nebulizer. The As detection limit for employed technique was 0.1 μg L<sup>−</sup>1. With the aim of understanding the metabolic pathway followed by this element once absorbed by the plant, we measured the amount of trivalent and pentavalent

As. Inorganic As species were extracted without the addition of hydrogen peroxide according to Rintala et al. [24] with some modifications. In brief, 200 mg of roots were homogenized and digested with 10 mL of a mixture of HNO3 (1%, v/v) and left to react overnight. The samples were subjected to microwave-assisted extraction according to the following program—3 min from 25 to 55 ◦C (step 1), 10 min at 55 ◦C (step 2), 2 min from 55 to 75 ◦C (step 3), 10 min at 75 ◦C (step 4), 2 min from 75 to 95 ◦C (step 5) and 30 min at 95 ◦C (step 6). Samples were then quantitatively transferred into tubes and centrifuged for 15 min at 10,000 rpm at 4 ◦C. The supernatant was filtered with a 0.22 μm PVDF filter. The concentration of As(III) was determined directly with an ICP-OES equipped with a hydride generation system. The total concentration of inorganic arsenic species [As(III)+As(V)] were obtained after reducing As(V) to As(III) through 5.0% (w/v) ascorbic acid and potassium iodide in hydrochloric acid, and the content of As(V) was calculated from the difference between total As concentration and that of As(III) [25].
