Accumulation of nutrients and agronomic performance of grafted seedless watermelon 1

equilibrated doses of fertilizers into the soil at sowing or as proper topdressing. The absorption of nutrients differs according to plant development, and is higher at flowering, setting and fruit growth. more plants Chemical fertilizing is one of the main inputs that increase watermelon production costs. Watermelon grafting with compatible rootstocks and a vigorous root system may increase the efficiency of nutrients absorption. This study aimed at determining the accumulation of dry matter and nutrients, as well as the agronomic performance of grafted watermelon. The triploid ‘Precious Petite’ watermelon was used as graft and the Line BGCIA941 ( Citrullus lanatus var. citroides ) as rootstock. The experimental design consisted of randomized blocks, in a split-plot, with six replicates. The ‘Precious Petite’ grafted onto Line BGCIA941, ungrafted ‘Precious Petite’ and ungrafted Line BGCIA941 were evaluated in the plot, while the subplot corresponded to the sampling time (0, 10, 20, 30, 40, 50, 60 and 70 days after transplanting). ‘Precious Petite’ grafted onto Line BGCIA 941 accumulated more macro and micronutrients than the ungrafted plants, increasing fruit yield, flesh firmness and rind thickness, but it did not affect the soluble solids contents. The decreasing order for nutrient accumulation in the grafted plants was N > K > Ca > Mg > P > S > Fe > Mn > Zn > B > Cu, whereas for ungrafted ‘Precious Petite’ it corresponded to N > K > Ca > Mg > P > S > Mn > Fe > B > Zn > Cu.


INTRODUCTION
Mineral nutrition is one of the factors that contribute directly to the yield and quality of watermelon fruits (Barros et al. 2012). Watermelon is relatively demanding in primary (nitrogen, phosphorus and potassium) and secondary macronutrients (calcium, magnesium and sulfur), and benefits from the micronutrients application. These nutritional demands may be suppressed by 1. Manuscript received in Jan./2016 and accepted for publication in Sep./2016Sep./ (http://dx.doi.org/10.1590Sep./ /1983 Chemical fertilizing is one of the main inputs that increase watermelon production costs. Watermelon grafting with compatible rootstocks and a vigorous root system may increase the efficiency of nutrients absorption. This study aimed at determining the accumulation of dry matter and nutrients, as well as the agronomic performance of grafted watermelon. The triploid 'Precious Petite' watermelon was used as graft and the Line BGCIA941 (Citrullus lanatus var. citroides) as rootstock. The experimental design consisted of randomized blocks, in a split-plot, with six replicates. The 'Precious Petite' grafted onto Line BGCIA941, ungrafted 'Precious Petite' and ungrafted Line BGCIA941 were evaluated in the plot, while the subplot corresponded to the sampling time (0,10,20,30,40,50,60 and 70 days after transplanting). 'Precious Petite' grafted onto Line BGCIA 941 accumulated more macro and micronutrients than the ungrafted plants, increasing fruit yield, flesh firmness and rind thickness, but it did not affect the soluble solids contents. The decreasing order for nutrient accumulation in the grafted plants was N > K > Ca > Mg > P > S > Fe > Mn > Zn > B > Cu, whereas for ungrafted 'Precious Petite' it corresponded to N > K > Ca > Mg > P > S > Mn > Fe > B > Zn > Cu.
The effects of rootstocks on the absorption, translocation and use of nutrients on plants have been documented in previous studies (Rouphael et al. 2008, Colla et al. 2010, Yetisir et al. 2013. The increase of nutrient absorption in grafted plants may be attributed mainly to the physical characteristics of the root system, such as the vertical and lateral development of the roots, which promote increased absorption of water and inorganic nutrients.
This study aimed at determining the effect of grafting on the accumulation of dry mass, absorption of nutrients and agronomic performance of triploid watermelon.

MATERIAL AND METHODS
The experiment was performed from June to August (2012), at the Bebedouro Experimental Field (9º9'S, 40º29'W and altitude of 365.5 m), which belongs to the Embrapa Semiárido, in Petrolina, Pernambuco State, Brazil.
The experimental design consisted of randomized blocks, in a 3 × 8 split-plot arrangement, with six replications. The following was evaluated in each plot: 'Precious Petite' cultivar grafted onto Line BGCIA 941, ungrafted 'Precious Petite' and Line BGCIA 941 rootstock. In the subplots, the following sampling periods were analyzed: 0, 10, 20, 30, 40, 50, 60 and 70 days after transplanting (DAT). The experimental unit consisted of 30 plants per plot.
The triploid 'Precious Petite' watermelon cultivar was used as graft, and the Line BGCIA 941 was adopted as rootstock, originating from the introduction of PI244019 (Citrullus lanatus var. citroides), which is conserved in the Active Cucurbitaceae Germplasm Bank of the Embrapa Semiárido.
The approach grafting method (Cushman 2006) was used at 9 days after sowing, when the seedlings showed the first clear leaf. After grafting, the seedlings were kept in a greenhouse (29 ºC ± 0.25 and 60 % RH ± 0.45). After 7 days of grafting, the root of the cultivar and the shoot of the rootstock were eliminated. On the eighth day after grafting, the surviving seedlings were transplanted to the field. The spacing adopted was 2.5 m × 0.5 m. Plants from the diploid 'Sugar Baby' cultivar were used in a 2:1 proportion, in alternate rows, as pollinators of the triploids.
The crop fertilization was carried out at 5 days prior to transplanting, based on the soil analysis and recommendations by Mendes et al. (2010), with 30 kg ha -1 of N, 120 kg ha -1 of P 2 O 5 and 60 kg ha -1 of K 2 O being applied, using a 06-24-12 commercial formulation of N, P and K, respectively. Topdressing fertilization was carried out with fertigation, using 50 kg ha -1 of N (calcium nitrate) and 30 kg ha -1 of K 2 O (potassium sulfate). The fertilization was performed three times a week up to 50 DAT. After the crop fertilization, the soil was covered with double-faced polyethylene (black within and white outside) to reduce weeding and maintain the soil humidity for a longer time.
After transplanting, the seedlings were placed in a low covered polypropylene tunnel, aiming at reducing the incidence of diseases at the initial development stage of the plants. At the beginning of flowering, the polypropylene was removed to allow pollination and the formation of fruits. Drip irrigation was performed and the amount of water applied daily was adjusted to the requirements of the crop, according to the climatic conditions, which were monitored by a meteorological station close to the experimental field.
Fruit harvest was carried out at 60 DAT for ungrafted 'Precious Petite' and at 70 DAT for the other plants. The incidence of wilt and death on ungrafted 'Precious Petite' plants was observed at 53 DAT, and Fusarium oxysporum was found to be responsible for it. However, the 'Precious Petite' plants grafted onto Line BGCIA 941 and the ungrafted Line BGCIA 941 survived until the end of the cycle, at 70 DAT. Therefore, F. oxysporum was the reason why the ungrafted treatment was harvested ten days earlier, if compared to the grafted treatment.
At transplanting, 20 plants per treatment were sampled to determine the concentration of nutrients at time '0'. Subsequently, four plants per replicate were sampled for the first two samplings (10 and 20 DAT), because of the small size of the plants. But, for the subsequent samplings (30, 40, 50, 60 and 70 DAT), two plants per replicate were used. The plants were harvested without roots (cut at the base of the stem), fractionated into leaves, stem + branches and fruits (when present), and washed once under running water with detergent and again with distilled water. The samples were dried in a forced-air circulation oven at 65 ºC, for 7 days. After drying, the vegetative matter was ground and digested to determine the concentration of macro and micronutrients.
The levels of N, P, K, Ca, Mg, S, B, Cu, Fe, Mn and Zn were determined according to Embrapa (2009). Subsequently, the accumulation curves, in relation to the dry mass, were estimated in g plant -1 . The dry matter of each plant (g plant -1 ) was determined by weighing the leaves, stems + branches and fruits (when present). The production of the grafted and ungrafted plants was assessed based on the number of fruits per plant and on the fruit weight (kg), being determined in tons per hectare, considering a plant density of 5,333.33. The following fruit quality characteristics of the grafted and ungrafted plants were examined: rind (exocarp + mesocarp) thickness (cm), by taking four measurements of the fruit (two on the side, one in the region of the stylar scar and one at the stem region); flesh firmness (N), by taking three readings of the fruit flesh with a manual penetrometer (FT327 model, 8.0 mm dial); and soluble solids (ºBrix), examined in five representative fruits of the plot by taking readings at three places in the fruit flesh, using a portable optical refractometer (RFC 28 model, at 62 %).
The data were submitted to analysis of variance and the means compared by the Tukey test, at 5 %. When the quantitative factor was significant (* p < 0.05, ** p < 0.01 and *** p < 0.001), the variables were adjusted to logistic functions of the model: Y = a/{1 + e [-b (day-c)] } (Verhulst 1838), where Y = mean value of the examined characteristic; a = asymptotic maximum (maximum quantity); b = mean increase rate of the accumulated quantity; day = time (days); c = time required to reach half of the maximum quantity.

RESULTS AND DISCUSSION
There was a significant effect of the interaction between the treatments (grafted/ungrafted) and days after transplanting, in the accumulation of dry mass and absorption of the nutrients N, P, Ca, Mg, S, B, Cu, Fe and Mn, in the total shoot (leaves + branches + fruits). The isolated effect of these factors was also significant. For the accumulation of K and Zn, although a significant effect of the factors was found independently, there was no interaction between them (Table 1).
Until 30 DAT, the 'Precious Petite' growth was similar to that observed for 'Precious Petite' grafted onto BGCIA 941, which accumulated 58.7 g plant -1 of dry mass. After this period, during the fruiting stage, the growth of 'Precious Petite' grafted onto BGCIA 941 (543.4 g plant -1 ) was greater than that for ungrafted 'Precious Petite', from 34 to 70 DAT ( Figure 1). The daily rate increased from 7 g plant -1 *, ** and ns : significant at p < 0.05 and p < 0.01 and non-significant, respectively.
An increase in the absorption rate of macronutrients was observed after the beginning of the fruiting stage (30 DAT), when the accumulation of dry mass increased (Figure 2). In the subsequent periods (at 30-60 DAT for 'Precious Petite', 30-70 DAT for 'Precious Petite' grafted onto Line BGCIA 941 and 20-70 DAT for Line BGCIA 941), there was a high demand for all macronutrients. Therefore, these are the ideal periods for topdressing fertilization.
It was observed that nitrogen accumulation in 'Precious Petite' and 'Precious Petite' grafted onto Line BGCIA 941 increased from 32 DAT (Figure 2a). From 38 to 70 DAT, the level of nitrogen in 'Precious Petite' grafted onto BGCIA 941 was higher than in ungrafted 'Precious Petite' and Line BGCIA 941, reaching 28 g plant -1 at 50-60 DAT. This trend was similar to that observed for the dry mass accumulation. The period of greatest absorption of N was 32-60 DAT for 'Precious Petite' and 20-70 DAT for BGCIA 941, when N reached 15 g plant -1 and 20.9 g plant -1 , respectively. The highest increase in nitrogen observed for 'Precious Petite' grafted onto Line BGCIA 941 was similar to that found by Colla et al. (2011), for 'Minirossa' watermelon grafted onto different rootstocks.
Phosphorus accumulated during the cycle for 'Precious Petite', 'Precious Petite' grafted onto Line BGCIA 941 and Line BGCIA 941 reached the greatest absorption rate at 33-60 DAT, 33-70 DAT and 30-70 DAT, respectively (Figure 2b). Phosphorus is the main nutrient involved in watermelon fruit yield, and the grafted plants in this study accumulated more phosphorus than the ungrafted ones. At 60 DAT, the accumulation of phosphorus in 'Precious Petite' grafted onto Line BGCIA 941 (1.8 g plant -1 ) was 38.4 % greater than that for 'Precious Petite'. These results differ from those observed for 'Ingrid' watermelon, which showed no difference in phosphorus accumulation between grafted and ungrafted plants (Rouphael et al. 2008). However, Yetisir et al. (2013) evaluated two watermelon cultivars grafted onto Lagenaria siceraria hybrids and Cucurbita spp. hybrids and found different amounts of phosphorus accumulation in watermelon. They also observed an influence of the rootstock and grafted cultivar on the accumulation of phosphorus.
The period of highest demand for potassium by 'Precious Petite' grafted onto Line BGCIA 941, 'Precious Petite' and Line BGCIA 941 occurred at 25-20 DAT, until the end of the cycle, when 11.1 g plant -1 , 7.3 g plant -1 and 9.0 g plant -1 were respectively accumulated (Figure 2c). At 40 DAT, 50 DAT and 60 DAT, the accumulation of potassium for 'Precious Petite' grafted onto Line BGCIA 941 was greater than for 'Precious Petite'. In watermelon, potassium plays an important role in fruit quality parameters such as color, size, acidity, resistance to transport and cracking of the rind, and the 'Precious Petite' grafted onto Line BGCIA 941 plants absorbed more potassium than ungrafted 'Precious Petite'. The potassium accumulation in the present Days after transplanting study differed from that for the Zaochunghongyu watermelon cultivar grafted onto Lagenaria siceraria and Cucurbita maxima x Cucurbita moschata hybrids, which showed no difference in potassium accumulation, in relation to the graft (Huang et al. 2012). From 38 DAT, 'Precious Petite' grafted onto Line BGCIA 941 absorbed more calcium than the other treatments (Figure 2d). The periods of increased calcium accumulation in 'Precious Petite', 'Precious Petite' grafted onto Line BGCIA 941 and Line BGCIA 941 were 28-60 DAT (10.1 g plant -1 ), 28-70 DAT (12.9 g plant -1 ) and 20-70 DAT (9.4 g plant -1 ), respectively. The degree of calcium accumulation by the plants is related to the blossom end rot of watermelon fruits. The increase in calcium absorption by the 'Precious Petite' grafted onto Line BGCIA 941 plants might reduce the incidence of this problem, in addition to conferring increased resistance to fruit mechanical damage. For 'Crimson Tide' and 'Crisby' watermelons, the calcium content was affected by the rootstock, regardless of the Lagenaria siceraria or Cucurbita spp. hybrids (Yetisir et al. 2013).
The period of greatest magnesium accumulation for 'Precious Petite' grafted onto Line BGCIA 941 and 'Precious Petite' occurred at 40-50 DAT, when 2.4 g plant -1 and 1.3 g plant -1 were respectively absorbed (Figure 2e). The absorption of magnesium was higher in 'Precious Petite' grafted onto Line BGCIA 941 at 30-45 DAT and for 'Precious Petite' at 30-40 DAT, with a daily accumulation rate of 0.23 g plant -1 and 0.30 g plant -1 , respectively. The maximum estimated accumulation of magnesium in 'Precious Petite' grafted onto Line BGCIA 941 surpassed the 'Precious Petite'. Similar results were found by Colla et al. (2010), who used a Cucurbita spp. hybrid as rootstock. However, the use of Lagenaria siceraria as a rootstock caused no difference between the grafted plants and the ungrafted 'Ingrid' watermelon.
The greatest accumulation of sulfur occurred from 30 DAT until the end of the cycle, for 'Precious Petite' grafted onto Line BGCIA 941 and 'Precious Petite', with a maximum of 1.5 g plant -1 and 0.8 g plant -1 and daily accumulation of 0.197 g plant -1 and 0.049 g plant -1 , respectively (Figure 2f). At 38 DAT and 60 DAT, the concentration of sulfur in 'Precious Petite' grafted onto Line BGCIA 941 was greater than that in 'Precious Petite', with 0.9 g plant -1 and 1.7 g plant -1 , respectively.
Until 38 DAT, the boron accumulation in 'Precious Petite' and 'Precious Petite' grafted onto Line BGCIA 941 was similar, but from 40 DAT and 60 DAT, the grafted plants accumulated more boron than the ungrafted ones. These results are similar to the findings for ungrafted 'Crimson Sweet' watermelon, which showed a maximum accumulation of 30 mg plant -1 of boron (Vidigal et al. 2009).
Until 70 DAT, 'Precious Petite' grafted onto Line BGCIA 941 accumulated 4.9 mg plant -1 of copper, with an absorption rate of 0.159 mg plant -1 day -1 (Figure 3b), whereas 'Precious Petite' and Line BGCIA 941 accumulated 2.8 mg plant -1 and 3.6 mg plant -1 of copper, respectively, with an absorption rate of 0.1 plant -1 day -1 . In contrast to the results of this study, the copper concentration for 'Ingrid' watermelon grafted onto Lagenaria siceraria and Cucurbita spp. hybrids was similar to that of the ungrafted plants (Colla et al. 2010).
At 60-70 DAT, the accumulation of iron in the total shoot was 91.0 mg plant -1 , 50.0 mg plant -1 and 91.8 mg plant -1 , respectively for 'Precious Petite' grafted onto Line BGCIA 941, 'Precious Petite' and Line BGCIA 941 (Figure 3c). In this period, the accumulation of iron in 'Precious Petite' grafted onto Line BGCIA 941 was greater than in 'Precious Petite', with 3.6 mg plant -1 at 30 DAT, 28.0 mg plant -1 at 40 DAT, 60 mg plant -1 at 50 DAT and 90 mg plant -1 at 60 DAT. For 'Crisby' watermelon, a reduction in the concentration of iron was found when it was grafted onto Lagenaria siceraria or Cucurbita spp. hybrid, whereas for 'Crimson Tide', the response varied according to the rootstock (Yetisir et al. 2013).
From 33 DAT to 55 DAT, 'Precious Petite' grafted onto Line BGCIA 941 showed a daily accumulation rate of manganese of 0.21 mg plant -1 , reaching a total accumulation in this period of 115 mg plant -1 (Figure 3d). For 'Precious Petite', a more rapid accumulation of manganese was found from 40 DAT to 60 DAT (47.1 mg plant -1 ), with 0.07 mg plant -1 day -1 . From 40 DAT, an increased manganese accumulation was observed for 'Precious Petite' grafted onto Line BGCIA 941, ranging from 15.8 mg plant -1 to 67.0 mg plant -1 (in relation to 'Precious Petite'), and from 53.3 mg plant -1 to 74.6 mg plant -1 (considering the Line BGCIA 941). In contrast to the data obtained for 'Precious Petite' grafted onto Line BGCIA 941, for 'Ingrid' watermelon, no difference was observed in the manganese accumulation, regarding the graft (Colla et al. 2010).
At 30-60 DAT, the accumulation of zinc by the total shoot of 'Precious Petite' grafted onto Line BGCIA 941 was greater than that for 'Precious Petite' (Figure 3e). The concentration of Zn in 'Precious Petite' grafted onto Line BGCIA 941 increased from 34 DAT to 50 DAT, reaching an accumulation rate of 25.3 mg plant -1 . The daily accumulation rate of Zn for 'Precious Petite' and the Line BGCIA 941 were similar, at 0.1 mg plant -1 day -1 , resulting in 16.4 mg plant -1 and 19.6 mg plant -1 at the end of the cycle, respectively. On the other hand, 'Precious Petite' grafted onto Line BGCIA 941 had a faster accumulation. For 'Crisby' and 'Crimson Tide' watermelons, a reduction in zinc accumulation was observed with the graft onto Lagenaria siceraria and Cucurbita spp. hybrids (Yetisir et al. 2013). According to Pulgar et al. (2000), grafted watermelon plants show an increased efficiency of zinc integration into nitrogenous compounds, forming chelates, which may promote an increase in the foliar biomass of grafted plants.
The graft did not alter the decreasing order of macronutrient accumulation by the shoot of the plants, which was the same for 'Precious Petite' grafted onto Line BGCIA 941, 'Precious Petite' and Line BGCIA 941: N > K > Ca > Mg > P > S. However, there was a change in the order of accumulation, especially for Mn and Fe, for 'Precious Petite' and 'Precious Petite' grafted onto Line BGCIA 941, and Zn and B for Line BGCIA 941 (   Similarly, for the Line BGCIA 941, Ca, Mg, B, Fe and Mn had higher concentrations in the vegetative parts, whereas N, P, K, S, Cu and Zn accumulated more in the fruits (Table 2). For grafted watermelon, it is recommended that chemical fertilization is reduced by one-half or two-thirds of the recommended dose for ungrafted plants (Lee & Oda 2010), in order to avoid excessive development of the shoots and promote a better distribution of these nutrients among the plant organs. For 'Crisby' and 'Crimson Tide' watermelons that were grafted onto Lagenaria siceraria and Cucurbita sp. hybrids, nitrogen accumulated more in the leaves than in the fruits, whereas phosphorus and zinc accumulated more in the fruits (Yetisir et al. 2013). Colla et al. (2006) found a greater accumulation of potassium in the vegetative tissues than in the fruits of the grafted plants.
There was a significant effect of the graft on the number of fruits, yield, fruit weight, rind thickness and flesh firmness, whereas the soluble solids were not affected (Table 3).
The number of fruits, total yield and marketable yield of 'Precious Petite' grafted onto Line BGCIA 941 were greater than that of 'Precious Petite' by 13.6 %, 54.9 % and 52.9 %, respectively (Table 4).
For 'Precious Petite' grafted onto Line BGCIA 941, a yield of 4.4 kg plant -1 was observed, which corresponds to an increase of 62.9 %, when compared to the yield of 'Precious Petite' (2.7 kg plant -1 ). The graft also promoted a 36.4 % reduction in the fruit weight of 'Precious Petite' (Table 4). An increase in production depending on the graft was found by Santos et al. (2014), for triploid 'RWT8154' watermelon grafted onto Citrullus sp., in agreement with the findings in this study. Similarly, the reduction in fruit weight observed for 'Precious Petite' grafted onto Line BGCIA 941 was also observed for 'Crimson Sweet' watermelon grafted onto Lagenaria ¹ Quantities of macro and micronutrients in the plant are expressed for grams and milligrams, respectively. siceraria, whereas 'Sugar Baby' showed a fruit weight increase with these rootstocks (Petropoulos et al. 2012).
In grafted plants, the rind thickness and flesh firmness of 'Precious Petite' increased by 66.7 % and 62.9 %, respectively (Table 4). However, the soluble solids were unaffected. An increase in rind thickness was found for the fruits of 'Precious Petite' grafted onto Line BGCIA 941 and also for 'Crimson Tide', 'Dumara' and 'Farao' submitted to grafting (Turhan et al. 2012). The increase in rind thickness and flesh firmness are attributed to increased calcium absorption (Rouphael et al. 2010 Table 3. Summary of analysis of variance for production and quality of grafted and ungrafted watermelon fruits.