Yield and physical characterization of Passiflora cincinnata in the Brazilian Savanna1

1 Received: Sep. 23, 2020. Accepted: Jan. 15, 2021. Published: Feb. 11, 2021. DOI: 10.1590/1983-40632021v5165795. 2 Universidade de Brasília, Brasília, DF, Brasil. E-mail/ORCID: anadabadia@yahoo.com.br/0000-0001-8401-0377. 3 Empresa Brasileira de Pesquisa Agropecuária (Embrapa Cerrados), Planaltina, DF, Brasil. E-mail/ORCID: ana-maria.costa@embrapa.br/0000-0003-4456-7802, fabio.faleiro@embrapa.br/0000-0001-8901-6428, juaci.malaquias@embrapa.br/0000-0001-6720-9624. 4 Empresa Brasileira de Pesquisa Agropecuária (Embrapa Semiárido), Petrolina, PE, Brasil. E-mail/ORCID: pinheiro.araujo@embrapa.br/0000-0002-5400-4686. For the Caatinga passion fruit (Passiflora cincinnata Mast.), there are no indications of varieties specific for the Brazilian Savanna conditions, as well as conduction techniques that make it possible to express a greater yield potential. This study aimed to evaluate the yield and quality of CPEF2220 and CBAF2334 populations in espalier and trellis conduction systems. A completely randomized design was used, in a 2 x 2 factorial arrangement, with three replications of four plants per plot. The fruit yield and physical characteristics were evaluated throughout the production cycle. The average yield was 8.0 kg plant-1 (3.5-14.9 kg plant-1) and the average number of fruits per plant was 139.1 (55.8-283.5), with average pulp yield of 29 %. There was a tendency to increase the number and yield of fruits for the CPEF2220 population conducted in the espalier system. The plant survival rate was 41 % (CPEF2220) and 87.5 % (CBAF2334), with a higher adaptation to the Brazilian Savanna conditions, while the CPEF2220 population presented a higher yield potential. Reductions in mass, longitudinal and equatorial diameter and fruit shape were observed throughout the harvest, with some exceptions for the espalier system, which showed more elongated fruits. The trellis system showed a greater yield potential for the parent populations of the passion fruit BRS Sertão Forte, for the study conditions.


INTRODUCTION
The Passiflora cincinnata Mast. species produces fruits of green color and is popularly known as Caatinga passion fruit, or bush passion fruit. It is native from the biomes Caatinga, Cerrado (Brazilian Savanna) and transition areas between the Brazilian For the Caatinga passion fruit (Passiflora cincinnata Mast.), there are no indications of varieties specific for the Brazilian Savanna conditions, as well as conduction techniques that make it possible to express a greater yield potential. This study aimed to evaluate the yield and quality of CPEF2220 and CBAF2334 populations in espalier and trellis conduction systems. A completely randomized design was used, in a 2 x 2 factorial arrangement, with three replications of four plants per plot. The fruit yield and physical characteristics were evaluated throughout the production cycle. The average yield was 8.0 kg plant -1 (3.5-14.9 kg plant -1 ) and the average number of fruits per plant was 139.1 (55.8-283.5), with average pulp yield of 29 %. There was a tendency to increase the number and yield of fruits for the CPEF2220 population conducted in the espalier system. The plant survival rate was 41 % (CPEF2220) and 87.5 % (CBAF2334), with a higher adaptation to the Brazilian Savanna conditions, while the CPEF2220 population presented a higher yield potential. Reductions in mass, longitudinal and equatorial diameter and fruit shape were observed throughout the harvest, with some exceptions for the espalier system, which showed more elongated fruits. The trellis system showed a greater yield potential for the parent populations of the passion fruit BRS Sertão Forte, for the study conditions. KEYWORDS: Caatinga passion fruit, tropical fruit, espalier and trellis conduction systems.
The BRS Sertão Forte (BRS SF) variety is available on the market, resulting from the crossing of two populations (CBAF2334 and CPEF2220) of plants originated and selected in the Brazilian semi-arid region. Under conditions of the Pernambuco state and in the Brazilian Savanna of the midwest plateau (highland), the variety may produce 18-30 t ha -1 year -1 , depending on the crop management conditions (Embrapa 2016).
Due to the climatic differences between the semi-arid region (Moura et al. 2007) and the Brazilian Savanna (Coutinho 2018), it is necessary to understand the productive performance of parents under specific conditions of the Brazilian Savanna for different conduction systems (espalier and trellis), what may contribute to understand the variety and establish strategies for cultivation recommendations.
The conduction of plants in a trellis system results in a higher fruit yield for the P. edulis species, if compared to the espalier one, whereas the P. setacea species yields about 13,754 kg ha -1 in the trellis and 10,492 kg ha -1 in the espalier systems (Komuro 2008, Costa et al. 2014. Although the trellis conduction system yields 30-40 % more, when compared to the wire espalier conduction system, it must be considered that the cost for implementing it is higher (Rigden 2011), since more support brackets are required to accommodate the main wire, as well as more time and labor. For P. cincinnata, there is lack of information regarding the conduction system, what is important for decisionmaking, due to the higher cost of implementation and maintenance when carried out in trellises, if compared to espaliers (Carvalho et al. 2015).
Thus, this study aimed to evaluate the yield and quality of CBAF2334 and CPEF2220 populations, which have the parental genotypes of the P. cincinnata cv. BRS SF, under cultivation conditions in the Brazilian Savanna, using the espalier and trellis conduction systems.

MATERIAL AND METHODS
Parental populations of the Passiflora cincinnata cv. BRS SF, CBAF2334 and CPEF2220, which are part of the Embrapa Semiárido germplasm bank, were evaluated. The populations were cultivated between April 2015 and July 2016, at the Embrapa Cerrados, in Planaltina, DF, Brazil (15º36'13.02"S, 47º43'17.34"W and average altitude of 1,050 m).
Pits with dimensions of 60 cm in diameter and 60 cm in depth were made with the aid of a motorized drill bit. Planting fertilization was carried out based on the soil analysis, using as a reference the fertilization for sour passion fruit: dolomitic limestone to raise the V to 50 % (Brasil & Nascimento 2010); 250 g pit -1 of P 2 O 5 (source: simple superphosphate); 100 g pit -1 of N (source: ammonium sulphate); 100 g pit -1 of K 2 O (source: potassium chloride); 100 g pit -1 of FTE BR12; and 10 L pit -1 of organic matter (source: chicken litter). The first cover fertilization was carried out at 60 days after planting, followed by others at 45-day intervals, using 100 g plant -1 (1:2 of potassium chloride and ammonium sulphate).
The crops were implemented in the field as shown in Figures 1A and 1B, with seedlings at the age of three months, on April 09, 2015, in the spacing between plants and between rows (2.5 m × 2.5 m for the espalier and 2.5 m × 5.0 m for the trellis conduction systems). The experimental design for the parents CPEF2220 and CBAF2334 was completely randomized, with three replications of four plants for each conduction system (trellis and espalier) ( Figure 1A).
For the study of fruit yield and physical characteristics carried out between November 2015 and July 2016, each population and its respective conduction system with three replications were considered, with 10 fruits per plot and all fruits of the plot being analyzed in the harvest of July 2016.
For the physical evaluations of fruits carried out between May and June 2016, a factorial arrangement (2 x 2) was considered, being two populations (CBAF2334 and CPEF2220) and two conduction systems (espalier and trellis), with three replications each and eight fruits analyzed per plot.
The climatic conditions of the cultivation period (April 2015 to July 2016) were expressed by an average temperature of 22.1 ºC, average relative humidity of 64.2 % and precipitation of 1,090 mm. The fruits of the plants from each plot were collected weekly from November 2015, the beginning of the first harvest. The collection took place after the abscission (after detaching from the plant, reaching physiological maturation), until the end of the e-ISSN 1983-4063 -www.agro.ufg.br/pat -Pesq. Agropec. Trop., Goiânia, v. 51, e65795, 2021 Yield and physical characterization of Passiflora cincinnata in the Brazilian Savanna material's life cycle (July 2016), with the last harvest comprising the fruits that fell on the soil and those that were trapped in the plant.
The analyzed data were: total number of fruits, obtained by counting the fruits per plant and per plot; longitudinal and equatorial length and diameter of the fruit, determined with a digital caliper (StainlessHardened™); and fruit mass, measured by a centesimal semi-analytical scale (OhausAdventurer™). The values for total mass and number of fruits were determined per plant. The mass and respective longitudinal (length) and equatorial (width) diameters were determined by analyzing 10 fruits from each plot, according to availability and collected weekly.
During the months of May and June 2016, of the ten fruits per plot, eight were evaluated for peel/skin thickness with a digital caliper (StainlessHardened™); peel and pulp mass (with and without seeds) obtained on a semi-analytical centesimal scale (OhausAdventurer™); pulp volume with and without seeds determined in a 100 mL graduated cylinder; and fruit shape obtained by the ratio between longitudinal (length) and equatorial diameter.
The seeds were detached from the pulp using a food blender with a blunt blade, and soon after they were separated from the pulp in a sieve. The seed fresh weight was determined by the difference in the pulp mass with and without seeds. The pulp yield was calculated in percentage values from the mass/mass ratio of the pulp without and with seeds per fruit, and the mass/mass ratio and volume/volume of the pulp without seeds per pulp with seeds.
Regression analyses were performed in each population and its respective conduction system to evaluate the monthly yield of fruits harvested on the ground between November 2015 and July 2016. For the physical characteristics of the fruits collected between May and June 2016, a 2 x 2 factorial scheme (populations x conduction systems) was used in a completely randomized design. The significance of the treatments was evaluated by variance analysis (Anova), the assumptions of normality for the residues were verified by the Shapiro-Wilk test (Miot 2017), and the homogeneity of variance by means of the Levene test (Levene 1960). For the comparison of means, the Tukey test was used, at 5 % of probability. All analyses were performed by the R statistical software, version 3.5.0 (R Core Team 2018).

RESULTS AND DISCUSSION
The total fruit yield of P. cincinnata (CPEF2220 and CBAF2334) conducted in the espalier and trellis systems ranged from 3.5 to 14.9 kg plant -1 , with mean value of 8.0 kg plant -1 . The linear regression analysis showed the existence of a different behavior in yield for the two populations in the two conduction systems (Figures 2A and 2B). There was a trend of linear increase in yield only for the CPEF2220 population conducted in the espalier system (Figure 2A).
A higher yield for the CPEF2220 population was observed in the trellis system (Figure 2A), which represented an average yield potential (125.7 %) higher than that for the espalier system. A similar behavior was observed in relation to the average number of fruits per plant, whose values varied from 55.8 to 283.5 fruits plant -1 , with an average of 139.1 fruits plant -1 . The CPEF2220 population showed a fruit yield 153.7 % higher than the CBAF2334 one.
Similarly to yield, a tendency to increase the number of fruits per plant ( Figure 2C) was observed only in the espalier system of the CPEF2220 population; while, in the trellis system of this population and in both systems for the CBAF2334 population, the same fact did not occur, indicating a random number of fruits per plant that suffer abscission along the production period and harvest. The obtained result was close to the average yield of 32 accessions of P. cincinnata cultivated in Petrolina (Pernambuco state, Brazil), in the espalier conduction system (Araújo et al. 2008), which was 15.88 kg plant -1 . Costa et al. (2014) also observed a higher yield for P. setacea (BRS PC) in the trellis conduction system, in relation to the espalier one, possibly due to the better distribution of the plant branches in the trellis, in relation to the espalier, and a consequent greater exposure to the sun and pollinators (Guimarães et al. 2013, Costa et al. 2014. The CPEF2220 and CBAF2334 parents showed a different behavior, concerning the ability to adapt to environmental conditions, regardless of the conduction system. The highest percentage of plant survival was found within the CBAF2334 population (87.5 %), in relation to the CPEF2220 one (41 %).
Individuals who increased the yield by up to three times with the density decrease, as well as others who did not respond to the decrease, were also verified, particularly within the CPEF2220 population.
There was a tendency for a significant decrease in the average fruit mass during the harvest (Figure 3), except for the CPEF2220 population, when carried out in the espalier system, which showed stability for fruit size from the beginning to the end of cultivation.
A trend towards a decrease in fruit mass with an increase in the number of fruits was observed for melon and sour passion fruit (Queiroga et al. 2008, Nogueira Filho et al. 2011, with the decrease being attributed to competition for plant reserves. The average fruit mass differed significantly only among the plant populations, with the CBAF2334 population presenting fruits of higher mass (Table 1). There was no significant interaction between conduction factors and plant populations.
The mean fruit longitudinal diameter of the two populations was 53 mm, while the equatorial diameter was 51 mm, varying throughout the fruiting period (Figure 4).
There was a significant trend towards a decrease in the longitudinal and equatorial diameters of the populations conducted in the trellis system ( Figure 4). This trend was not observed for the populations when conducted in the espalier system, except for the longitudinal diameter of the CBAF2334 population. Queiroga et al. (2008) also verified a reduction in the length and width of melon fruits throughout the crop cycle, which was attributed to a higher yield in the number of fruits, a fact that may also have occurred with the P. cincinnata fruits (Figures 2C and 2D).
The relation between the longitudinal and equatorial diameters indicates the shape of the fruit, and, the closer the value to 1, the more rounded the fruit will be (Jesus et al. 2015). The fruits of the whole trial showed values that varied from 0.95 to 1.14, indicating a shape close to the rounded one. However, the fruits of the CBAF2334 population proved to be significantly more elongated (Table 1).
Significant differences in the fruit shape were observed between the espalier and trellis conduction  (Table 1), with little evident practical effect. There was a tendency for a significant reduction in the fruit shape, except for the CPEF2220 population conducted in the espalier system, which tended towards a constant shape ( Figure 5). Komuro (2008) evaluating P. edulis and Silva et al. (2004) P. alata also observed differences in the fruit size, when using conduits in a vertical espalier and T espalier (which resembles the trellis because it is positioned horizontally). Variations in the size and shape of P. edulis fruits, as a function of the growing environment, were also observed by Scorza et al. (2017). The fruit shape and dimensions are regulated by genes belonging to the MAD-box complex and play an important role during the reproductive development (Cutri & Dornelas 2012). Scorza et al. (2017) studied two genes of this complex in P. edulis and found that PeFUL, FRUITFULL (FULL) homologue, presented a broad pattern of expression in vegetative and reproductive tissues, suggesting that these genes may also be related to the development of P. cincinnata fruits.
There was no interaction between the conduction systems and plant population, nor any  Yield and physical characterization of Passiflora cincinnata in the Brazilian Savanna significant variation, in relation to the seed mass (Table 2). Significant differences, however, were observed in the peel mass and thickness and in the seedless pulp mass and volume between the CPEF2220 and CBAF2334 plant populations ( Table 2). The CPEF2220 population plants showed peel with lower mass and thickness, while fruits of the CBAF2334 population had a greater mass and volume of seedless pulp, indicating that the two genotypes have beneficial characteristics for consumption in retail and industry. The volume of pulp with seeds for the CBAF2334 population (Table 2) presented a value close to that described by Magalhães (2010) (47.1 mL). The peel thickness of the CPEF2220 population recorded values that can be classified as very thin, and CBAF2334 as thin (Jesus et al. 2015). The significant differences observed in the mass and volume of the seedless pulp (Table 2) between the plant populations were correlated with the greater fruit mass of the CBAF2334 population (Table 1).
There was no interaction or significant differences between the P. cincinnata populations and the conduction systems, in relation to pulp yield. The yield of the seedless pulp mass, in relation to the fruit mass, was 26.63-31.37 %, with an average value of 29 %, while the mass of pulp with seeds, in relation to the fruit mass, was 63.04-66.58 %, with an average value of 64.81 %. The mass yield of pulp without seeds, in relation to the pulp with seeds, was 40.18-47.29 %, with an average value of 43.73 %, while the volume yield of pulp without seeds, in relation to the pulp with seeds, was 48.88-49.10 %, with an average of 44.99 %. Lessa (2011) observed, for P. cincinnata and the BRS SF cultivar, a pulp yield of 31.88 % and 28.70 %, respectively, values close to those found in the present study. These values are also close to those found for the P. edulis BRS Sol do Cerrado cultivar (Tupinambá et al. 2012), which presented an average pulp yield with seeds of 31.45 %, in relation to the fruit.   1. The trellis conduction system enabled the greatest yield potential of the Passiflora cincinnata populations in the conditions of the Brazilian Savanna, with a greater yield potential for the CPEF2220 population and a higher survival rate for the CBAF2334 population; 2. The P. cincinnata fruits showed characteristics of very thin skin/peel and pulp yield compatible with the fruits of P. edulis. The trellis conduction system reduced the average value of mass, shape, longitudinal (length) and equatorial (width) diameter of the fruits throughout the fruiting period; 3. The espalier conduction system influenced the fruit shape, which was more elongated, regardless of the evaluated population.