**2. Material and Methods**

Two experiments were conducted on tomatoes in spring–summer season in commercial greenhouses in the Split area (Mediterranean area of Croatia). The first experiment used plants grown on one stem, and in the second experiment two-stem plants were used.

#### *2.1. Experiment I: Single Stem Plants (2016)*

The first experiment was established in an unheated greenhouse in the Trogir (43◦31 N, 16◦15 E) used for intensive vegetable production for many years. The soil type was an alkaline clay with 8.09 pH

(H2O), 7.51 pH (KCl), 4.4% soil organic matter and 120 mg of available P2O5, and 48 mg K2O/100 g of soil. The greenhouse had side ventilation and the roof was 3 m high.

Tomato (*Solanum lycopersicum* L., cv. Clarabella F1, Rijk Zwaan, The Netherlands) plants were either self-grafted or grafted onto the rootstocks "Emperador" (*S. lycopersicum* × *S. habrochaites*, Rijk Zwaan) or "Maxifort" (*S. lycopersicum* × *S. habrochaites*, De Ruiter Seeds, Amsterdam North, The Netherlands). Both rootstocks, as noted by the seed companies, have high to medium vigor and are resistant to *Fusarium*, *Verticillium* and ToMV.

The scion seeds were sown on 17 January 2016 and rootstock seeds on 20 January 2016 in polystyrene plug trays with cell volume of 40 mL in an organic substrate (Brill Type 4, Brill Substrate, Georgsdorf, Germany). As scion and rootstocks have variable growth vigor and to ensure optimum stem diameter between scion and rootstock seedlings at grafting time the scion seeds were sown 3 days earlier than the rootstock. The trays with sown seeds were put in a heated greenhouse (day/night 25 ◦C).

The cv. 'Clarabella' seedlings were self-grafted and grafted onto both rootstocks at 25 days after sowing using the "splice grafting" method. Grafted seedlings were maintained under reduced light conditions (10% of the daily light intensity) at a relative humidity above 95% and temperature from 22 ◦C to 25 ◦C until callus formation. After callus formation, the seedlings were maintained as standard tomato transplant. Seedlings were grown in research greenhouse at the Institute for Adriatic Crops (Split, Croatia).

Tomato seedlings with four to five true leaves were transplanted 65 days after sowing (24 March 2016) in a two-row system (90 cm apart) with rows 60 cm apart with plants were spaced 50 cm in each row, for a total of 2.7 plants/m2. The plants were drip irrigated with drippers (pressure-compensating emitters) set in opposite lines of row plants (15 cm from plants). After transplanting, plants were irrigated per the standard practice in the area. During the trial, plants were fertilized twice with Mg, as other nutrients had high available soil concentrations, as confirmed by N and K petiole sap analysis during growth phases.

Three irrigation treatments were started 30 days after transplanting: full irrigation (FI), deficit irrigation (DI) and partial-root zone drying (PRD). The soil moisture content was measured by tensiometers (Blumat digital, Leingarten, Germany) which were laboratory calibrated for conducted soil. The tensiometers were placed at 25–30 cm depth. In FI treatment, plants were watered to a soil moisture content of 65–75% of field capacity. DI treatments received 50% water used in FI by using drippers with half capacity than in FI. With PRD, half of the root system was irrigated at 65–75% of field water capacity (FWC), while other half of the roots were dried until soil moisture reached 35–40% of field capacity and then irrigation was shifted between two sides of the root system. PRD also received 50% water from FI. Different irrigation treatments were divided by placing PE folia to the depth of 45–50 cm to stop horizontal water movement. Taking into account water supplied before start of irrigation treatment, in total DI and PRD got 60% water supplied to FI that received 214 L per plant.

The plant height and the number of leaves (longer than 5 cm) were determined for 7 weeks after transplanting. Harvest started 70 days after transplanting (DAT) and lasted for 55 days—including 12 harvests of fruits as they matured (light red color) The average fruit weight and fruit number were recorded. The first four harvests were calculated as early yield. On the day of the last harvest, the aboveground parts of the subsample plants (3 plants per treatment) were removed, divided into leaves, stems and green fruits and weighed for fresh biomass (FM). After measuring the leaf area with leaf area meter LI-3000 (LI-COR, Bad Homburg, Germany), samples were put into an oven and dried at 70 ◦C to constant weight to obtain the DM. The yield divided by supplied water was used to find the yield WUE (WUEy).

#### *2.2. Experiment II: Double Stem Plants (2017)*

The greenhouse in the split was used for the second experiment (43◦30 N, 16◦30 E). The soil type was a clay loam with 8.71 pH (H2O), 7.46 pH (KCl), 2.9% soil organic matter, 16.5% high active lime and 27 mg of available P2O5 and 31 mg K2O/100 g of soil. Same rootstocks were used as in Experiment I, and scion cultivar was Attiya (Rijk Zwaan, De Lier, the Netherlands) due to resistance to TSWV (Tomato spotted wilt virus) that influenced growth and yield in previous years on different tomato cultivars in used greenhouse. The scion and rootstocks seeds were sown on 19 January 2016 and 23 January 2016, respectively. Grafting was done on 17 February 2016; all other procedures for seedling production were done as in previous year. In this experiment, two types of seedlings were produced: self-grafted Attiya grown on one stem (ATT) and plants grown on two stems: self-grafted Attiya (AT), Attiya grafted on Emperador (EM) and Attiya grafted on Maxifort (MX), which in total gave four plant types. Two-stemmed seedlings were formed from side-shoots of cotyledons.

The tomato seedlings were transplanted 55 days after sowing (16 March 2016) in two systems: one stem seedlings as in previous years in a two-row system and two-stem seedlings in a one-row system with rows 120 cm apart. In each row, the plants were spaced 50 cm, which in both systems gave 2.7 stems/m2.

Irrigation treatments started 50 DAT and included FI, DI and PRD. In this year, the soil moisture content was controlled by Maxi Rain soil moisture sensors (Elektronik Jeske, Windorf, Germany), which were set up to open electromagnetic valve when soil moisture was lower than 65% field water capacity and was irrigated until reached 80% FWC. DI treatment received 60% water supplied to FI using drippers with lower capacity. To better measure water needs in the PRD, this treatment had its own sensors for controlling irrigation and switching sides. The PRD used 50% of the water used in FI. In total—including water applied before irrigation treatment start—the DI used 70% and PRD 65% water of the FI. The FI in total received 233 L/plant, DI 170 L/plant and PRD 153 L/plant. House-made lysimeters (60 cm × 100 cm × 5 cm) were put below every irrigation treatment at depth of 60 cm to control possible leaching if plants were over-irrigated.

Plants were fertilized by irrigation system with N, K and Mg, depending on growth phases and plant needs. Plant height and leaf number were measured each week. Harvest started at 80 DAT (7 June) and lasted 45 days, consisting of 11 harvests. At the end, plants were divided to determine biomass partitioning, as in previous year.

Leaf nutrient concentrations were determined in the youngest fully developed leaves after the leaves were dried at 70 ◦C and then ground. The micro-Kjeldahl digestion method (Kjeltec System 1026, Tecator, Höganas, Sweden) was used to measure total leaf N concentration. Dry ash of grounded samples from a muffle furnace were dissolved in 2 mL HCl to extract the P, K, Ca and Mg. The K concentration was measured using a flame photometer (Model 410, Sherwood Scientific, Cambridge, UK). The vanadate-molybdate yellow color method using a UV-visible spectrophotometer was used to determine the P concentration (Cary 50 Scan, Varian, Palo Alto, CA, USA) at 420 nm. The Ca and Mg in solution were determined by atomic absorption spectrometry (SpectrAA 220, Varian, Palo Alto, CA, USA).

The quality parameters of fruits from each treatment were analyzed in the second experiment. For the tomato juice, the total soluble solids (TSS) content was determined by a DR 201–95 refractometer (Kruss optronic, Hamburg, Germany) and expressed in Brix at 20 ◦C. Acidity was determined by juice titration with 0.1-M NaOH was used for determination of acidity and results were expressed as citric acid. Gas-exchange parameters were measured using LI-6400 infrared gas analyzer (LI-COR, Inc., Lincoln, NE, USA) in youngest fully expanded leaves. Measurements were performed on six leaves per treatment 20 days after different irrigation techniques were applied in whole experiment. Measurements were conducted under constant light (PAR 750 μmol m−<sup>2</sup> s<sup>−</sup>1) and CO2 concentration (400 μmol mol<sup>−</sup>1). The environmental conditions in the greenhouse ranged from 22 ◦C to 33 ◦C for air temperature and from 33% to 42% for relative humidity (RH). The greenhouse light conditions (PAR) ranged from 300 to 1100 μmol m−<sup>2</sup> s−1. The transpiration rate (E) and photosynthetic rate (A) were determined from gas exchange measurements and were used to determine the photosynthetic/instantaneous water-use efficiency (PWUE) as the ratio between A and E [19].

The experiments were set up in a randomized block design, consisting of three replications. Each treatment (irrigation × plant type) was comprised of 12 plants. The data were evaluated by

ANOVA and when F-tests were significant, the means of the main factors (rootstock/plant type and irrigation technique) and their interactions were compared using the least significant difference test at *p* ≤ 0.05. The data were statistically analyzed using StatView ver. 5.0 (SAS Institute, Cary, CA, USA).
