Grafting height does not affect Fusarium wilt control or horticultural performance of Passiflora gibertii N.E.Br. rootstock

The influence of the grafting height (5, 10, 20 and 30 cm above the root collar) of P. edulis on P. gibertii was evaluated on the incidence of Fusarium wilt and horticultural performance. Plants of P. gibertii grafted on P. edulis and non-grafted plants of both species were also studied. In addition, histopathological studies were also performed on the roots of non-grafted P. edulis collected at three severity stages of Fusarium wilt. In greenhouse, the graft take was inversely related to the grafting height in general. In the field conditions, the plant growth of P. gibertii grafted on P. edulis was superior to its reciprocal grafting, even though the former combination was susceptible to Fop . Plants of P. edulis grafted on P. gibertii at all grafting heights did not present symptoms of Fop, and the number of fruit yield and quality were equivalent, but plant growth was decreased in relation to the non-grafted plants. Starch depletion in the root system of P. edulis was directly related to the severity of the Fusarium wilt. P. gibertii was confirmed as a Fusarium wilt resistant rootstock of P. edulis , with minimal influence of the grafting height for the control of the disease.


INTRODUCTION
Soil-borne diseases are the main causes of economic losses in the yellow passion fruit (Passiflora edulis Sim) cultivation (Ortiz and Hoyos-Carvajal 2016, Morgado et al. 2017). Fusarium wilt, caused by Fusarium oxysporum f. sp. passiflorae (Fop), is considered the most important because it attacks the vascular system causing sudden wilting and early death of the plant (McGovern 2015). Fop has been decreasing the productivity and lifespan of P. edulis crop worldwide (Fischer and Rezende 2008).
The grafting onto resistant rootstocks is used for the control of soil-borne diseases in several crops (Prunier et al. 1999, Pavlou et al. 2002, Sayler et al. 2002 either anticipates the production (Castle 1995, Karlidag et al. 2016. The grafting of the yellow passion fruit on several resistant wild species was studied as an alternative for the production in 2 LUCAS K.S. LIMA et al. areas with a history of Fop (Cavichioli et al. 2011a, Lima et al. 2017. Passiflora gibertii N.E. Br. was highlighted with additional desirable traits, such as rapid plant growth, high seed germination and good anatomical compatibility for grafting with P. edulis (Cavichioli et al. 2011a, Yockteng et al. 2011, Lima et al. 2017. Studies with passion fruit, in general, do not make inferences of the effect of grafting height on the horticultural performance and incidence of Fop, even though this influence is reported for other grafted crops and diseases (Prunier et al. 1999, Sayler et al. 2002, Chalise et al. 2013, Yazdani et al. 2016, Karlidag et al. 2016. Nakasone and Paull (1998) report that higher grafting height of the purple passion fruit on the yellow one is associated to a decreased incidence of Fusarium wilt. Cleft grafting of P. edulis on P. gibertii at 6 to 8 cm above the soil level resulted in high plant survival to Fusarium wilt in São Paulo State, Brazil (Cavichioli et al. 2011a). However, infected plants of this combination were observed in Bahia State, Brazil, which was attributed to the low height in hypocotyledonar grafting (Santos et al. 2016).
In this work, we evaluated the influence of the grafting height of P. edulis on P. gibertii on the incidence of Fusarium wilt and horticultural performance. In addition, the grafting of P. gibertii on P. edulis and histopathological aspects of the latter's root system were investigated to understand this pathogen-host relationship.

LOCATION OF EXPERIMENT AND PLANT MATERIAL
Experimental works were carried out in the municipality of Cruz das Almas, Bahia State,Brazil (12° 39' 25'' S,39° 07' 27'' W,222 m. a. s. l.). Plantlets were grown in greenhouse from January to April 2016, at 28±2 °C and relative humidity of 60%. Later, the plantlets were transplanted into a field with a high incidence of Fusarium wilt (Santos et al. 2016) and cultivated from April 2016 to August 2017. During the field evaluation period, the average air temperature was 25.4 °C, with rainfall of 1047 mm and relative humidity of 77%.
The scion variety was BRS Rubi do Cerrado, a commercial yellow passion fruit hybrid (P. edulis), and the wild species P. gibertii (BGP008 accession) was used as the rootstock. Non-grafted seedlings of both species were controls. Reciprocal grafting was also performed to evaluate if the P. gibertii scion would induce some resistance to Fop in the rootstock of P. edulis ( Figure 1).

EVALUATION IN GREENHOUSE
The experimental design in the greenhouse was completely randomized blocks in a factorial scheme (2 x 4 + 2), with two scion-rootstock combinations (P. edulis/P. gibertii and P. gibertii/P. edulis) and four grafting heights (5, 10, 20 and 30 cm above the root collar, Figure 1), plus two additional treatments (non-grafted seedlings of each species), distributed in four replications of 10 plants per plot.
The rootstock seedlings grafted at 20 and 30 cm height were sown 30 days before those grafted at 5 and 10 cm height. This procedure was performed so that the grafts were performed at the same time at all grafting heights studied. The seeds were planted in trays using vermiculite potting mix. Fifteen days after emergence, the rootstock seedlings were transplanted into polyethylene bags (22 x 12 cm) containing a mixture of humus, vermiculite, coconut fiber and commercial potting mix Vivato ® (2:1:1:1, v:v).
At 45 days after transplanting, the cleft grafting method was performed and the rootstocks were cut with scissors at heights of 5, 10, 20 and 30 cm. At the cutting site, a longitudinal cut of 1.0 cm was opened with a scalpel blade. In the open cut, the scion with about 4.0 cm long, obtained from the meristematic apex of seedlings from passion GRAFTING HEIGHT IN PASSION FRUIT 3 same of the greenhouse, except that six plants were used in the plot. The soil of the experimental area is classified as allic distrocohesive yellow latosol, with the following chemical characteristics (0-20 cm): pH (water) 6.5, phosphorus (P) 60.85 mg dm -3 , potassium (K) 0.20, calcium (Ca) 2.26, magnesium (Mg) 0.99, aluminum (Al) 0.0, sodium (Na) 0.09, hydrogen + aluminum (H + Al) 1.25, base saturation (SB) 3.24, cation exchange capacity (CTC) 4.79 cmol c dm -3 , base saturation percentage (V%) 73.65%, and organic matter (OM) 13.2 g kg -1 . The plants were trained by single-wire trellises with 2.0 m height and spacing between plants was 1.8 m x 2.0 m. Drip irrigation was carried out three times a week with a duration of 30 min each, with an average water volume of 10 L per plant per irrigation. The plants were fertilized every 20 days for the first six months with 30 g of ammonium sulphate per plant (afterward increased to 80 g) fruit seeds 'BRS Rubi do Cerrado', was inserted. The base of the scion was cut into wedge-shaped double bevel and its leaves were cut in 1/3 of its area. Subsequently, the graft was inserted in the rootstock carefully, in order to coincide with the cambium tissues; the grafting region was protected by clip grafted (Figure 1).
The variables studied were graft length (GL, in cm) measured from the graft region to the tip of the plant, diameter of the scion stem (DC, in mm) measured with a pachymeter at 2 cm above the grafting region, and graft take (%) at 60 days after grafting (DAG).

EVALUATION IN FIELD CONDITIONS
At 60 DAG, all the grafted plantlets and seedlings were transplanted into the field to evaluate the incidence of Fusarium wilt and the horticultural performance. The experimental design was the The plant growth index (GRI) was estimated up to 220 days after transplanting as established by Jesus et al. (2016). The number of fruits per plant and crop yield (kg ha -1 ) were also measured in the first production cycle. To evaluate fruit quality, five ripe fruits were randomly collected per plot at peak plant production during the months of November and December 2016. Fruit quality was studied only for the plants of the P. edulis/P. gibertii combination and the non-grafted seedlings of P. edulis. The physical and chemical characteristics measured were: fruit mass (g), length and diameter (cm), skin thickness (mm) and mass (g), pulp mass (g), soluble solids content (SS, in °Brix), titratable acidity (TA) expressed in mg citric acid per 100 mL juice, as determined through titration with NaOH at 0.1 mol L -1 , and the maturity index ( Root samples were fixed, embedded, sliced in sections and visualized according to the method described by Lima et al. (2017). Three slides containing 10 sections were prepared for each treatment, totaling 30 units. To determine the presence of starch, sections containing 5 μm thick root segments were stained with Lugol's iodine for 5 min for observation of cell structure with dark blue or brown color (Berlyn and Miksche 1976).
For fluorescence microscopy, the root segments were stained with Lugol's iodine for 3 min, then washed with tap water and stained with 1% aniline for 8 min and finally in Lugol again for 30 s and assembled with tap water (Kraus and Arduin 1997). The blue aniline dye produces a blue color in callose tissue and Lugol acts on the cell walls, giving gray and yellowish coloration to lignified ones. To verify the presence of callose in the Fop-infected yellow passion fruit root system, a fluorescence microscope with ultraviolet filter was used (Axioskop2, Carl Zeiss, Jena, Germany). The slides were also analyzed under a light microscope.
Hyphae of the pathogen were visualized using fresh root fragments 2-4 cm long to evaluate the colonization of host tissues by Fop, using the clearing and staining technique (Phillips and Haymann 1970).

STATISTICAL ANALYSES
Data were submitted to analysis of variance and the means were compared by the Tukey test at 5% probability. When necessary, data were transformed by to satisfy the premises of normal distribution. Survival analysis was performed using the nonparametric Kaplan-Meier (KM) curve (Kaplan and Maier 1958). In this study, the event of interest was the death of the plant caused by the Fusarium wilt, which was recorded every two days based on the typical visual symptoms ( Figure 2). The differences between survival curves in the different combinations of scionrootstock and grafting heights were tested using the

EVALUATION IN GREENHOUSE
Biometric variables and percentage of graft take presented significant differences (p ≤ 0.05%) in relation to the scion-rootstock combinations and grafting heights ( Figure 3). There was interaction between the scion-rootstock vs. grafting height factors for the variables graft length and stem diameter (Figures 3a, d). P. gibertii/P. edulis presented higher graft length than the P. edulis/P. gibertii combination at all grafting heights except for 30 cm (Figure 3a). Plants with grafts at 5 cm showed the least growth, especially when P. edulis was grafted on P. gibertii (Figure 3a). In relation to non-grafted plants, P. gibertii presented greater plant height with 76.0 cm, superior to P. edulis with 41.0 cm (Figure 3b). Mean plant height of non-grafted plants (58.4 cm), was higher than average graft length of all scionrootstock combinations with 45.8 cm (Figure 3c).
For the stem diameter, there were different responses between the combinations and grafting heights ( Figure 3d). P. gibertii/P. edulis presented the largest stem diameters at all grafting heights except 5 cm, at which the reciprocal combination (P. edulis/P. gibertii) had wider stem diameter ( Figure 3d). The stem diameter of non-grafted P. edulis was 3.66 mm, 27% thicker than P. gibertii with 2.68 mm (Figure 3e). The grafted and nongrafted plants did not differ significantly for stem diameter (Figure 3f).
Regarding the graft take of P. edulis/P. gibertii, the grafting height at 10 cm provided the highest percentage (92%), although it did not differ significantly from plants grafted at 5 cm, with 87% ( Figure 3g). In contrast, the lowest percentages were observed at 20 cm and 30 cm, both with 62% ( Figure 3g). In general, the non-grafted plants showed higher plant survival (93%) than the grafted plants with 75% ( Figure 3i). This behavior was not maintained among non-graft plants (Figure 3h).

EVALUATION IN FIELD CONDITIONS
In field conditions there was no effect of grafting height on the plant growth index (GRI 7.57) ( Figure  4a). When the P. gibertii scion was grafted on the P. edulis rootstock, a higher GRI 6.72 (p = 0.01) occurred, 20% higher than the inverse combination ( Figure 4b). Regarding the non-grafted plants, opposite behavior was observed, with higher GRI for P. edulis (9.53) than P. gibertii (5.89) ( Figure  4c). There were no differences between the grafted and non-grafted GRI plants of 7.57 and 7.71, respectively (Figure 4d).
The number of fruit per plant and the productivity were not influenced by the evaluated grafting heights, and there was no difference in these variables between the grafted plants and nongrafted P. edulis plants (Figures 4e-f).
The Peto and Peto's Wilcoxon, log-rank and Cox tests indicated that the survival curves generated by the Kaplan-Meier nonparametric method differed significantly (p < 0.0001) between the non-grafted P. edulis and P. gibertii seedlings and the combinations P. gibertii/P. edulis and P. edulis/P. gibertii (Figure 5a).
The use of P. gibertii/P. edulis grafting affected the susceptibility to Fusarium wilt of P. gibertii, which is resistant to the disease. The manifestation of symptoms by P. gibertii/P. edulis plants started at 150 days after planting (DAP), and, at the end of the study, only 32% of plants were still alive (Figure 5a).
The non-grafted seedlings of P. edulis were more susceptible to Fusarium wilt, with the onset  (Figure 5c). Survival to Fop of the P. gibertii/P. edulis combination was significantly lower independent of the grafting height used, and no variation was observed between the survival curves by the three tests (Peto and Peto's Wilcoxon, p = 0.31964, Cox, p = 0.45898, and log-rank, p = 0.16047) ( Figure  5d). Lower dispersions attributed to Fop were obtained at 5 and 20 cm grafting heights, with the first quartile between 220-260 and 180-210 days at these grafting heights, respectively, when compared to grafts at 10 and 30 cm (Figure 5e).  The mean survival time was lower (175 days) for the grafting performed at 5 cm ( Figure 5f) and the intensification of mortality occurred from 250 days after planting (Figure 5d). Based on the physicochemical characteristics of the fruits of P. edulis grafted on P. gibertii, there were no differences (p ≤ 0.05) between the grafting heights for all the evaluated characteristics ( Table  I). The grafting height did not influence the average number of fruit per plant when compared to that of non-grafted P. edulis (Figure 4e).

HISTOPATHOLOGICAL ANALYSIS OF ROOTS
The fluorescence microscopy images of the P. edulis seedling (non-grafted) plants without visual symptoms of Fusarium wilt revealed the presence of fibers in the cortical parenchyma of the root with light staining and deposition of starch in the same region extending to the xylem vessels (Figures 6a,  e). In the first wilting stage, there was a marked reduction in the presence of starch in comparison to the control, in addition to a smaller volume of fibers (Figures 6b, f). In the second wilt stage, complete absence of starch was observed in both the cortex and xylem (Figures 6c, g), and in the third wilt stage, this behavior was maintained, with further deterioration of vascular tissues (Figures 6d, h). No callus deposition was observed in the xylem vessels in any of the wilt stages, indicating that the species P. edulis does not have this metabolite compound, which constitutes a physical barrier that prevents the colonization of the pathogen (Figures 6f-h).
The presence of starch in the cortical parenchyma (Figures 6i, m) was identified in the transversal sections of the root of healthy plants stained with Lugol's iodine, while in the fungus-infected plants, low starch concentration was observed in the first wilt stage (Figures 6j, n) in comparison to the control, and complete absence of this polysaccharide in the other wilt stages (Figures 6k, l, o, p).
With the root clearing technique, it was possible to identify the presence of hyphae inside cells in secondary roots in the first wilt stage ( Figure  6r) and the presence of chlamydospores in the last wilt stage (Figure 6t). No pathogen structures were observed in control plants, which had healthy roots with well-delimited cells (Figure 6q).

DISCUSSION
The definition of grafting techniques for yellow passion fruit is important to obtain vigorous plants associated with resistance to the main soil-borne diseases that constrain cultivation in several regions of Brazil. In the present study, results indicated that as the grafting height increases, there is a reduction in the graft take in the nursery.
In field conditions, no differences in the plant growth were observed regarding grafting heights, On the other hand, the growth rate of P. gibertii when grafted on P. edulis was higher than the inverse combination. These results are very interesting, since they indicate that the reduction in the vegetative vigor of P. edulis when grafted on P. gibertii is associated with the developmental limitation of the root system of the rootstock and not with the vascular connection process. This assumption is supported by the larger GRI of P. gibertii/P. edulis combination in relation to the non-grafted P. gibertii plants. P. edulis grafted on P. gibertii showed lower growth speed compared to P. edulis. A lower root dry mass in P. edulis/P. gibertii was observed in potted plants in greenhouse in relation to the self-graft of P. edulis (Morgado et al. 2015). These results corroborate the size of the root system of P. gibertii as the limiting factor for the development of P. edulis scion. The root system of the rootstock is of fundamental importance, since the water flow to the scion controls many plant processes, such as growth, mineral nutrition, photosynthesis, transpiration and, consequently, productivity (Martínez-Ballesta et al. 2010, Gambetta et al. 2012.
All plants of P. edulis/P. gibertii and seedlings of P. gibertii survived throughout the evaluation period in an area with a Fop history, regardless the grafting height, indicating that the resistance of P. gibertii occurs at the root level. Therefore, the lower survival of P. edulis/P. gibertii reported by Santos et al. (2016) may be associated with the crop management, climatic conditions or genetic variation within the rootstock seedlings, among other factors. There are few studies on the effect of the grafting height on the incidence of soil-borne diseases of fruit crops. The incidence of Pseudomonas sp. was reduced in apricot and Prune due to grafting height of 180 cm in relation to 30 cm above the root collar (Prunier et al. 1999, Sayler et al. 2002. This range of grafting height is obviously of no practical use in vine crops such as passion fruit. On the other hand, the use of P. gibertii as a scion grafted on P. edulis resulted in high mortality of plants associated with Fop throughout the same period. These results confirm the existence of inhibitory compounds of infection and/or colonization of the pathogen in the roots of P. gibertii, and that there is no flow of these substances from the scion to the roots of the susceptible species, P. edulis. Recent studies indicate the movement of miRNA via phloem from shoot to root in vines, which, however, is strongly dependent on the environment (Yang et al. 2015).
The maintenance of the yield and quality of the yellow passion fruit in grafted plants is also fundamental for the acceptance of rootstocks by growers. The results obtained showed that there were no changes in the physical and chemical attributes as a function of the grafting heights tested in relation to the non-grafted P. edulis plants. The number of fruit and productivity were also not influenced by the grafting height or in comparison with non-grafted seedlings of P. edulis. The productivity in the first production cycle in the grafted plants observed in this study was below (39%) the national average which is of 14.1 t ha -1 (IBGE 2017), which is calculated on the basis of at least two production cycles over the year. Nevertheless, the average life expectancy of nongrafted P. edulis was only 168 days, indicating that the number of live plants in the second production cycle will be less than 40%. Cavichioli et al. (2011a) found a production of 3,517 kg ha -1 in P. edulis/P. gibertii from December 2006 to February 2007, which did not differ from self-grafted P. edulis.
Grafting is an important alternative for living with Fusarium wilt until obtaining a variety of P. edulis with resistance to the pathogen. The grafting should only be used in areas infested with Fop, because although there is no compromise of the physical and chemical characteristics of the fruits, the vigor and productivity of grafted plants, when compared with those of seedling plants, might be reduced in areas without the disease. Cavichioli et al. (2011b), when evaluating the quality of passion fruit fruits from plants of P. edulis grafted on three rootstocks (P. edulis, P. alata and P. gibertii), verified that grafted plants produced fruits satisfying marketing standards, despite being shorter in length than the non-grafted plants. No significant rootstock effects on antioxidants contents or fruit quality of P. edulis were observed in the graft combinations of P. edulis/P. mucronata and P. edulis/P. gibertii The histopathological evaluations of roots of non-grafted P. edulis plants at different Fusarium wilt stages demonstrated that as disease progresses, there is a significant reduction of starch in the roots, and the presence of hyphae and chlamydospores, especially in the last wilt stage. The fiber distribution and lignification of the cortical walls in the secondary roots was low, both in control plants and in the different wilt stages, with reduction in the fluorescence of the fibers in the infected plants.
These physical structures represent an important barrier to pathogen infection and the their fragility may be associated with the susceptibility of yellow passion fruit to Fop, since the cell wall is considered the first barrier to penetration or colonization of the pathogen. Furthermore, changes in this membrane can occur that make it impossible for the pathogen to colonize the plant (Underwood 2012, Miedes et al. 2014. The available information in the literature on starch dynamics in the plant-pathogen interaction is scarce, and the few studies that have been carried out show that pathogen-affected plants present a reduction in the deposition of this polysaccharide in roots (Keunen et al. 2013, Manila andNelson 2014). In the healthy plants, greater starch accumulation was observed in the root cortex, and as the wilt stage intensified, the starch concentration declined or was too low to detect. This reduction may also be associated with the stage of wilt observed in plants regardless of the pathogen, since studies show that starch degradation is associated with osmotic adjustment to water deficit (Ponce et al. 2008). Nevertheless, the most important anatomical and histopathological changes due to Fop infection were the concentration of starch, hyphae and chlamydospores as a function of the wilt stage.

CONCLUSION
Graft take of P. edulis on P. gibertii in greenhouse was higher at 5 and 10 cm height. In field conditions, the grafted P. gibertii/P. edulis plants presented higher growth than the reciprocal grafting and non-grafted P. gibertii. However, they were susceptible to Fop, indicating that resistance occurs at the root level. Grafting height from 5 cm up to 30 cm above the root collar did not influence the Fusarium wilt incidence of P. edulis grafted on P. gibertii, thus this combination is considered resistant to Fop. Fruit yield and quality were also not influenced by the grafting height or in relation to the P. edulis seedlings. Based on the fluorescence and histopathological analyses of the roots, it was possible to identify changes in starch concentration and presence of hyphae and chlamydospores as a function of the Fusarium wilt stage in non-grafted P. edulis plants. These phenomena are possibly related to the infection and colonization processes in this pathosystem.