Grafting onto an Appropriate Rootstock Reduces the Impact on Yield and Quality of Controlled Deficit Irrigated Pepper Crops

In this study, hybrid pepper rootstock NIBER® is tested for its ability to overcome water stress situations under soil conditions. The impact of deficit irrigation (DI) on yield and fruit quality, irrigation water use efficiency is evaluated, and consequently, the agronomic impact of employing water-stress tolerant rootstock is compared to ungrafted pepper plants. For this purpose, plants of the California-type sweet pepper ‘Maestral F1’ grafted onto NIBER® underwent a sustained DI regime during seasons 2018 and 2019 and were compared to their respective controls. Plants were drip-fertirrigated, and volumetric soil water content was continuously monitored by capacitance sensors. Gas exchange and leaf water potential measurements were taken early in the morning and midday 58, 79, and 114 days after transplanting. Plant and fruit dry biomass, marketable quality, blossom-end rot incidence and harvest index were also determined. For consecutive years, our results confirmed that grafting a pepper cultivar onto an appropriate rootstock (NIBER® in this case) as part of a DI strategy can overcome the negative effects of sustained water stress conditions. The plant biomass production and fruit yields of grafted plants were less affected by DI due to less sensitivity to water stress. This can be attributed to a less marked reduction in shoot dry weight in the grafted plants, which allowed greater whole photosynthesis by maintaining sink activity compared to ungrafted plants.


Introduction
Sweet pepper (Capsicum annuum L.) is an important vegetable crop around the world, with production at approximately 36 Mt obtained from 2 million ha. Currently, the leading sweet pepper-producing countries are China, Mexico, and Turkey, while Spain, Italy, and Romania are the main producers in Europe [1].
Drought stress is an important limiting factor for vegetable crop production. Water is becoming increasingly scarce worldwide, which seriously affects agricultural production, especially in arid and semiarid areas [2,3], due to increased global temperature and evapotranspiration (ET) and less

Experimental Site
The experiment was conducted during two consecutive growing seasons (GS; 2018 and 2019) at the Instituto Valenciano de Investigaciones Agrarias (IVIA) Research Institute in Moncada (Valencia, Spain; latitude: 39.589517, longitude: -0.395550, elevation: 37 m). The climate is the Mediterranean, distinguished by warm wet winters and hot, dry summers. Temperature and relative humidity values along the experimental seasons are presented in Figure S1. Trials were conducted in an unheated plastic greenhouse (30 m long × 7.5 m wide) that was E-W oriented, and passively ventilated by opening side panels and roof vents. The soil composition within a 20-centimeter depth was sandy clay loam (68% sand, 11% clay, 21% silt) and contained 0.61% organic matter, 0.051% total N, 8 mg kg −1 of P, 301 mg kg −1 of K, and 2.87 meq 100 g −1 of assimilable Mg. Soil electrical conductivity (1:5) was 0.29 dS m −1 , and pH was 8.1. Irrigation was applied with a drip system with one single line per plant row and one emitter every 0.2 m with a 2.5 L h −1 discharge rate. Irrigation water had an electrical conductivity of 0.3 dS m −1 , 16.5 mg L −1 of NO 3 − content, very low concentrations of other macroand micronutrients, and a pH of 8.1. Nutrients were applied during both crop seasons through the irrigation system according to the following nutritional plan and in line with Maroto [30]: 170 kg ha −1 of N; 40 kg ha −1 of P 2 O 5 ; 200 kg ha −1 of K 2 O.
Seedlings were transplanted on 27 March 2018 and 2 April 2019 inside the plastic greenhouse in raised beds spaced 1.10 m apart with one plant row per ridge and 0.5 m spacing between plants within rows (1.8 plants m −2 ). The raised part of the raised bed was 0.2 m wide, 6 m long, and 0.10 m high. Plants were horizontally supported by two nylon guide cords laid parallel to both sides of the plant line, as described by Maroto [30], and no pruning was performed.
The experiments were laid out to be complete randomized with three replicates. Each replicate consisted of 13 plants.

Irrigation Strategies
Two irrigation strategies (IS) were adopted: Irrigation with 100% and 50% of the estimated crop water requirements (control (C) and DI, respectively). Irrigation during the first four weeks after plantation was applied to assure the correct plant establishment. Irrigation management consisted of fixed 30-minute irrigation events, applied as often as necessary, depending on the VSWC and irrigation strategy. Crop irrigation water requirements were obtained by estimating crop evapotranspiration (ET c ) by the K c -ET o method [31]. The reference evapotranspiration (ET o ; in mm day −1 ) was calculated by the calibrated radiation method, which is often used in the greenhouse industry in SE Spain [32]. This method requires daily solar radiation outside greenhouses and cladding greenhouse transmissivity (σ, in %). Solar radiation was obtained from an agroclimatic weather station located near the greenhouse, while transmissivity was calculated by using the radiation data collected before the crop period with a CMP3 pyranometer (Kipp and Zonen, Delft, The Netherlands), whose value was 43% during both seasons. The crop coefficient (K c ) values were those determined by Orgaz et al. [33] for sweet pepper. Initially, K c took a value of 0.2 and rose to 1.4 at harvest. The applied irrigation water, and its distribution according to growth stages, are presented in Table 1. For the environmental conditions during the experimental period, the ET o calculated from outdoor solar radiation was 261.4 mm in 2018 and 265.1 mm in 2019. When applying the proposed K c , the estimated water requirements were 230.5 mm in 2018 and 228.4 mm in 2019. The total applied water amount in the control strategy was 216.3 mm and 206.8 mm, respectively (Table 1), which means that the irrigation enforced in the control strategy was around 94% and 91% of the ET c value in 2018 and 2019, respectively. The DI (V-DI and V/N-DI) received 58% and 56% of the control doses (V-C and V/N-C) in 2018 and 2019, respectively.

Soil Moisture
The volumetric soil water content (VSWC; m 3 m -3 ) was continuously monitored by ECH 2 O EC-5 capacitance sensors connected to an Em50 data logger using the ECH 2 O Utility software (Decagon Devices, Inc., Pullman, WA, USA). VSWC values were used directly because factory calibration provides ± 3% accuracy for mineral soils. One sensor per replicate was placed below the dripline at a 20-centimeter depth, where the maximum root density was located and was equidistant between two adjacent emitters. The VSWC was measured and stored every 30 min, and its variation was used to determine in situ field capacity (FC) [34,35]. The permanent wilting point (PWP) was determined by using Richards Plates (Set for pF-determination; Eijkelkamp, Giesbeek, The Netherlands) [36]. To compare VSWC between irrigation strategies and growing seasons, their values are presented as available water content (AWC; %) before each irrigation event, and were determined as reported by Fernández et al. [37]: considering the subscripts a, FC and PWP correspond to the actual (before irrigation), field capacity, and permanent wilting point soil water content, respectively. When the VSWC a was greater than the VSWC FC , the last one was considered, given that AWC cannot be greater than 100%. The values for the VSWC at FC and PWP were determined for the different GS and IS and are shown in Table 2.

Physiological Measurements
Physiological measurements were taken during GS 2019. The gas exchange measurements were taken early in the morning (9.30 am to 10.30 am GMT) and at midday (1 pm to 2 pm) with three plants per replicate (9 per treatment) after 58, 79, and 114 days after transplanting (DAT). The net CO 2 assimilation rate (A N , µmol CO 2 m −2 s −1 ), stomatal conductance (g s , mol H 2 O m −2 s −1 ), and transpiration rate (E leaf , mmol H 2 O m −2 s −1 ) were determined on fully expanded leaves (3rd-4th leaf from the apex) in the steady-state under saturating light conditions (1000 µmol m −2 s −1 ) and with 400 ppm CO 2 by an LI-6400 infrared gas analyzer (LI-COR, Nebraska, USA) at 24 ± 2 • C and 65 ± 10% relative humidity.
The average values of leaf vapor pressure deficit (considering 58, 79, and 114 DAT) were 1.68 ± 0.22 kPa and 2.50 ± 0.25 kPa at early morning and midday, respectively. Parameters A N /g s and A N /E leaf were calculated as intrinsic water efficiency and instantaneous water use efficiency, respectively.
The predawn (5.00 am to 6.00 am) and midday leaf water potential (1 pm to 2 pm) (Ψ predawn and Ψ leaf , respectively) were determined by a Schölander-type pressure chamber (model 3000; Soil Moisture Equipment Corp., Santa Barbara, CA, USA). Determinations were made on fully expanded leaves in an identical physiological state as that used in the gas exchange measurements of three plants per replicate (9 per treatment) after 58, 79, and 114 DAT.
Plant conductance of the hydraulic water flow from root systems to leaves (whole-plant hydraulic conductivity, Kh plant ) was calculated by analogy to Ohm's law [38]: where E leaf-md is the maximum transpiration rate in the leaf measure in LICOR-6400 from 1 pm to 2 pm by assuming that Ψ soil = Ψ predawn . Kh plant was measured at 58, 79, and 114 DAT.

Biomass and Fruit Yield
Harvests consisted of four passes between mid-June and the end of July in both years. The commercial production quality was evaluated following the criteria described by the European Regulations [39], insofar as the yield was partitioned into three categories: «Extra» Class, Class I (together hereafter referred to as marketable yield; MY) and Nonmarketable, the fruits which, given their defects, mainly due to blossom-end rot (BER), did not match the former categories.
Biomass was analyzed at the end of crop cycles. The aboveground plant part was partitioned into two parts and separately analyzed: Vegetative, including shoots with all their leaves (hereafter referred to as shoot dry weight (SDW)), and fruit (including all the fruit of all the passes of the harvests). The sum of both terms was the total dry weight (TDW). Each part was dried at 65 • C in a forced-air oven (Oven 100-800, Memmert, Germany) until constant weight and weighed on an analytical scale, which measures dry weights. The harvest index (HI) was determined as the ratio of the total yield to the total aboveground biomass (TDW) on a dry mass basis [37].

Statistical Analysis
The biomass and fruit yield data were subjected to an ANOVA analysis, including the three factors: IS, plant type (PT), and GS. The physiological parameters were separately analyzed for each measurement time (58, 79, and 114 DAT) as a multiway ANOVA (Statgraphics Centurion for Windows, Statistical Graphics Corp.). With these parameters, after verifying the significance of the interaction for each variable (data are not shown), a one-way ANOVA was performed by joining the plant combination and treatment for each studied time. Means were compared by the Fisher's least significance difference (LSD test) at p < 0.05. No significant differences were found among the replicates for each measured parameter. in 2019], which was lower than those reached in the control treatment. In 2018, this drop in AWC was more marked for V/N-DI than for V-DI. From early June, soil moisture also decreased in the control treatment in the two PT and during both GS. It is noteworthy that in 2019, some problems were encountered with the irrigation system, which involved an unplanned lack of irrigation for two days. Consequently, the AWC in the control treatments dropped below 50%, and rose to previous levels when irrigation was reestablished. It is also noteworthy that the standard error was generally lower for DI than for the control plants as soil moisture remained at constant values.

Soil Moisture
decreased in the control treatment in the two PT and during both GS. It is noteworthy that in 2019, some problems were encountered with the irrigation system, which involved an unplanned lack of irrigation for two days. Consequently, the AWC in the control treatments dropped below 50%, and rose to previous levels when irrigation was reestablished. It is also noteworthy that the standard error was generally lower for DI than for the control plants as soil moisture remained at constant values.

Plant Water Relations
The predawn leaf water potential (Ψpredawn) values ( Figure 2A) after 58 DAT were not significantly different between the control and DI treatments. At 79 DAT and at the end of the

Plant Water Relations
The predawn leaf water potential (Ψ predawn ) values ( Figure 2A) after 58 DAT were not significantly different between the control and DI treatments. At 79 DAT and at the end of the experiment (114 DAT), the V-C and V/N-C values were higher (with no significant differences between them) compared to V/N-DI and V-DI, of which the latter had the lowest Ψ predawn values. The midday leaf water potential Ψ leaf values progressively lowered during the experiment as plant size enlarged in all the IS ( Figure 2B). The application of the DI strategy significantly decreased Ψ leaf in the DI plants, but only at the end of the experiment (114 DAT), with no differences appearing between the grafted and ungrafted plants.
Whole plant hydraulic conductivity (Kh plant ) ( Figure 2C) was not constant during the experiment. At 58 DAT, values ranged between 7 and 10 mmol H 2 O m −2 MPa −1 s −1 , and the lowest values were for the V-DI plants, and the highest ones were measured for the V-C plants, whereas the grafted plants had

Photosynthetic Parameters
The net CO 2 assimilation rate (A N ) measured early in the morning ( Figure 3A) showed no significant differences between plants and treatments during the crop cycle. Nevertheless, the A N values obtained at midday ( Figure 3B) were grouped for the significant differences in the control and DI plants at 58 DAT and 79 DAT. The lowest values were for the DI plants at both these times. At 114 DAT, the maximal A N was observed in the V-C plants followed by V/N-C, V-DI, and V/N-DI with the lowest values.
The transpiration rate (E leaf ) early in the morning ( Figure 3C) increased from 58 DAT to 79 DAT, but with no significant differences between treatments at both DATs. At the end of the experiment, the DI strategy lowered the E leaf values in the V and V/N plants compared to the control strategy. At midday, the E leaf values of the control plants were higher than the values obtained early in the morning ( Figure 3C,D). The measurements taken at 58 DAT and 79 DAT did not show any changes in the E leaf of the V-C and V/N-C plants, but the E leaf values were higher in the V-C plants than in the V/N-C plants at the end of the experiment. The application of the DI strategy led to equal E leaf values for both the V and V/N plants at 58 DAT ( Figure 3D). At 79 DAT, the DI strategy only reduced E leaf in the V-DI plants. At the end of the experiment, the differences in E leaf between IS were more evident, with the highest values for V-C and the lowest ones for V/N-DI.
Stomatal conductance (g s ) was measured early in the morning ( Figure 3E) and only showed significant differences between IS at the end of the experiment, with similar reductions in the V and V/N plants compared to the control. The influence of IS on g s was more evident at midday ( Figure 3F). Although the measurements taken at 58 DAT and 79 DAT did not indicate any changes in g s of the V-C and V/N-C plants, g s was higher in the V-C than the V/N-C plants at the end of the experiment. At 58 DAT, the reduction in g s accomplished by the DI strategy was equal for the V and V/N plants ( Figure 3F). At 79 DAT, the DI strategy only reduced g s in the V-DI plants. At the end of the experiment, the differences in g s between IS were more evident, with the highest values for V-C and lowest ones for V/N-DI. Instantaneous water use efficiency (AN/Eleaf) progressively decreased early in the morning for all the plant combinations during the experiment, with minimum values at 114 DAT ( Figure 4A). There were no significant differences between IS at 58 DAT and 79 DAT, whereas AN/Eleaf increased in the DI plants (V and V/N) at 114 DAT compared to the control. A similar response occurred at midday ( Figure 4B). Regarding intrinsic water use efficiency (AN/gs), a different shape behavior was observed during the experiment to the values measured early in the morning ( Figure 4C) and at midday ( Figure  4D). Early in the morning ( Figure 4C Instantaneous water use efficiency (A N /E leaf ) progressively decreased early in the morning for all the plant combinations during the experiment, with minimum values at 114 DAT ( Figure 4A). There were no significant differences between IS at 58 DAT and 79 DAT, whereas A N /E leaf increased in the DI plants (V and V/N) at 114 DAT compared to the control. A similar response occurred at midday ( Figure 4B). Regarding intrinsic water use efficiency (A N /g s ), a different shape behavior was observed during the experiment to the values measured early in the morning ( Figure 4C) and at midday ( Figure 4D). Early in the morning ( Figure 4C), it was only at the end of the experiment when the A N /g s values exhibited significant differences, with the highest values for the DI plants (V-DI and V/N-DI).

Plant Biomass and Fruit Yield
Shoot dry weight (SDW) was significantly affected by IS (P ≤ 0.01; Table 3), with a 20% decrease for the stressed plants versus the control ones, but the IS × PT interaction for this parameter was statistically significant (P ≤ 0.05; Table 3) as the differences in SDW between DI and C were small (down to 50%) in the V/N plants in relation to V ( Figure 5). Total dry weight (TDW) was also significantly affected by IS (P ≤ 0.01; Table 3), with an average 22% decrease for the DI plants in relation to the control for both seasons. No significant IS × PT interaction was detected. The harvest index (HI) was affected by neither any analyzed factor nor their interaction for any season.

Plant Biomass and Fruit Yield
Shoot dry weight (SDW) was significantly affected by IS (p ≤ 0.01; Table 3), with a 20% decrease for the stressed plants versus the control ones, but the IS × PT interaction for this parameter was statistically significant (p ≤ 0.05; Table 3) as the differences in SDW between DI and C were small (down to 50%) in the V/N plants in relation to V ( Figure 5). Total dry weight (TDW) was also significantly affected by IS (p ≤ 0.01; Table 3), with an average 22% decrease for the DI plants in relation to the control for both seasons. No significant IS × PT interaction was detected. The harvest index (HI) was affected by neither any analyzed factor nor their interaction for any season.  The mean values followed by the different lowercase letters in each column indicate significant differences at P ≤ 0.05 using the LSD test. ** and * indicate significance at P ≤ 0.01 and P ≤ 0.05, respectively, n.s. no significant difference. (+) Standard deviation, calculated as the square root of the residual sum of squares. df degrees of freedom. The marketable yield (MY) and «Extra» class yield were also significantly affected by IS (P ≤ 0.01; Table 4) and by the IS × PT interaction (P ≤ 0.05; Table 4). Deficit irrigation reduced yields in relation to the control, down to 24% and 31% for the MY and «Extra» class yield, respectively. In both cases The marketable yield (MY) and «Extra» class yield were also significantly affected by IS (p ≤ 0.01; Table 4) and by the IS × PT interaction (p ≤ 0.05; Table 4). Deficit irrigation reduced yields in relation to the control, down to 24% and 31% for the MY and «Extra» class yield, respectively. In both cases for the DI plants, higher yields were maintained in the V/N plants compared to V ( Figure 6). Class I yield was affected by neither any analyzed factor nor their interaction.   The «Extra» class number of fruits per plant and average fruit weight were only affected by IS (Table 5) as the DI values lowered by 25% and 8%, respectively, in relation to the control. As previously mentioned, in 2018, no fruits were affected by BER, and were not affect by any analyzed factor in 2019 (Table S1). The mean values followed by the different lowercase letters in each column indicate significant differences at p ≤ 0.05 using the LSD test. ** and * indicate significance at p ≤ 0.01 and p ≤ 0.05, respectively, n.s. no significant difference. (+) Standard deviation, calculated as the square root of the residual sum of squares. df degrees of freedom.

Discussion
For two consecutive years, our results demonstrated that grafting a pepper cultivar onto the new hybrid rootstock, NIBER ® , can overcome the negative effects of sustained water stress in biomass and yield terms.
The water applied during both GS was similar because the meteorological conditions were comparable that, in turn, led to equivalent water requirements. Harvesting is, as usual, the longest growth stage, and it presents the maximum daily water requirements. This means having to apply the largest amount of water, which represents 58% of the total amount.
For both GS and IS, AWC maintained close to 100% from planting to the establishment, which indicates that the VSWC came close to FC. The reduction of AWC after the establishment was more marked for V/N-DI than for V-DI, in 2018, and was probably related to the fact that water uptake was greater in the grafted plants. Afterward, the AWC in the DI strategies dropped to 40% and 50% in 2018 and 2019, respectively. These values were lower than 55%, the value established by Fernández et al. [37] as the threshold AWC, below which pepper ET c lowers in response to soil water deficit. Therefore, it can be stated that DI plants were water-stressed, and as the AWC for both plant types was similar during each GS, the stress level was similar for both PT. So, it can be hypothesized that the higher yield obtained in V/N in relation to V could be due to the grafted plants' better response to water stress conditions.
In fact, DI reduced both shoot and total dry weight by an average 20-22% for the DI plants (V/N-DI and V-DI) in relation to the control for both seasons, which is a similar decrease to that observed by López-Marín et al. [24], and comes close to the reduced plant yields reported by several studies for pepper crops [40,41] and tomato [42]. What is more, the NIBER ® rootstock had a beneficial effect on MY and «Extra» Class by maintaining a bigger yield than the ungrafted plants under stress conditions. These results, together with those previously obtained with water stress tolerant rootstocks [29], confirm our hypothesis on NIBER ® water stress tolerance, now under field conditions, and monitoring soil water content.
Fereres and Soriano [18] pointed out that when water stress increases in severity terms at a sustained DI, with biomass production losses over 40%, the HI can be affected in many crops. However, in the present study, the HI was not affected by either IS or PT, which indicates that yield was lowered by DI to a similar extent as vegetative biomass. Similar results have been obtained by Abdelkhalik et al. [41], who observed differences for the HI between treatments, but only when sever DI was applied, and these authors did not observe any differences for a moderate deficit.
The reduction in SDW between the deficit and control irrigation regimes in the grafted plants was much less marked than in the ungrafted plants, around 56% less. This indicates the ability of some rootstocks to overcome the effects of water stress, as previously found by López-Marín et al. [24]. According to Lee et al. [43], the root system of selected rootstocks is usually much larger and more vigorous than that of cultivars, which allows them to more efficiently absorb water and nutrients compared to ungrafted plants. The NIBER ® rootstock has shown better root development in several experiments (Reference [44] and unpublished data). As stated before, water content monitoring in the soil in our experiments affirms the better response of the NIBER ® rootstock to water stress in biomass formation and yield terms.
We also tested the physiological processes associated with grafting and DI by focusing on water and photosynthetic relations to explain the obtained agronomic results. Under well-irrigated conditions, plant water status (Ψ predawn and Ψ leaf ) was apparently similar between the V/N and V plants. Similar results have been reported in other studies with mini-watermelon, where the leaf water potential in both ungrafted and grafted plants showed comparable values upon optimal irrigation [23]. However, the use of pepper plants grafted onto a drought-tolerant rootstock had a significant influence on plant water transport capacity compared to the ungrafted plants. The hydraulic conductance of the entire plant in the control treatment under high atmospheric demand conditions (at noon) was lower in V/N-C, and was accompanied by lower g s values. Changes in hydraulic properties and stomatal behavior enable grafted plants to better regulate the water used by the plant by reducing the water lost by transpiration, which improves leaf water use efficiency (A N /E leaf and A N /g s ). This can be attributed to the differences in the root system's water absorption capacity, as found in grafted melon plants [45].
We observed that DI strongly influenced the plant water status both in the grafted and ungrafted plants, especially at the end of the experiment (114 DAT), which coincides with the fruit ripening stage, and a similar plant water stress level was obtained (based on Ψ leaf ≈ -1.8 MPa). The DI effects were also reflected in lower Kh plant , which means that plant water transport capacity diminished compared to well-irrigated plants, and was similarly affected in the V and V/N plants. Changes in plant water relations affected g s in both plant groups, with maximum stomatal regulation at midday. Stomatal limitation partially reduced the plant's water use, but also lowered the CO 2 assimilation rate similarly in both the V and V/N plants. In both plant groups, water stress had less impact on photosynthesis than in g s and E leaf , which indicates that stomatal closure occurred earlier than CO 2 fixation [26,46], and consequently and similarly enhanced water use efficiency (A N /E leaf and A N /g s ).
The physiological alterations generated by DI significantly reduced yield and biomass production in both plant groups (V and V/N). This reduction has been associated mainly with changes in all the photosynthetic parameters and water relations, as observed in different crops like mini-watermelon [23], tomato [47] or pepper [24,27,29] when comparing grafted vs. ungrafted plants. Nevertheless, the highest fruit yields and SDW under DI were recorded in the grafted plants (V/N-DI) compared to the V-DI plants, despite no significant differences being observed between them from the physiological point of view. The greater SDW reduction in the V-DI plants (25.7%) versus V/N-DI (14.5%), could lead to a more marked decrease in the intercepted radiation in the V-DI plants and could, thus, reduce the whole canopy photosynthesis in relation to V/N-DI. Although root biomass was not monitored in this experiment, SDW decreased in V-DI plants, and this could be explained by differences in biomass partitioning between the aerial part and roots. On the other hand, the capacity to maintain sink activity in the vegetative V/N-DI parts could be an adaptive advantage that leads to more shoot growth and bigger yields through regulating source-sink relations [24] with DI. Other mechanisms not associated to maintenance photosynthetic activity have been described to explain the better biomass and yield performance in grafted plants under DI as osmotic adjustment [48], modulation of antioxidant defenses [49] or hormonal signaling [50].

Conclusions
These results confirm that using appropriate rootstocks in pepper crops is a good strategy to better tolerate sustained DI, which promotes water stress in pepper, compared to ungrafted plants. This is a consequence of an adaptation morphology and physiology being capable of maintaining photosynthesis levels, and consequently, biomass production. They also confirm that hybrid NIBER ® is an interesting rootstock for this purpose.