Performance of 'Valência' sweet orange grafted on trifoliate orange hybrid rootstocks

– The objective of this work was to evaluate tree size, production, and fruit quality of 'Valência' sweet orange ( Citrus sinensis ) grafted on various trifoliate orange rootstocks, in order to select genotypes with a high performance. Twenty rootstock genotypes were evaluated, including trifoliate orange hybrids with mandarin (citrandarins) and with grapefruit (citrumelos), as well as 'Rangpur' lime. The experiment was implemented in the northwestern region of the state of Paraná, Brazil, in a 6.0×2.5 m spacing, in a sandy soil under subtropical and rainfed conditions. The statistical model used was the randomized complete block design with four replicates and four trees per plot. Tree size and fruit yield and quality were analyzed during three consecutive harvest seasons. Data were subjected to the analysis of variance, and means were grouped by the Scott-Knott test. Principal component analysis and agglomerative hierarchical clustering were also carried out. F.80-18, F.80-5, and F.80-3 citrumelos and IPEACS-239 citrandarin are adequate rootstock options for 'Valência' sweet orange, with dwarf trees and a high production efficiency. These rootstocks, except F.80-18, are also adequate options to obtain oranges with good industrial properties.


Introduction
'Valência' sweet orange [Citrus sinensis (L.) Osbeck] is one of the most cultivated varieties worldwide, with the characteristics -high fruit production and quality -required by the juice industry and by the fresh fruit market (Saunt, 2000;Bastos et al., 2014).It is graft-compatible with various rootstocks, including 'Rangpur' lime (Citrus limonia Osbeck) and 'Swingle' citrumelo [Citrus paradisi MacFad.× Poncirus trifoliata (L.) Raf.], the most used in Brazil, as well as with some trifoliate orange hybrids, which are a new generation of rootstocks with potential for use in the citrus industry (Simonetti et al., 2015).
Citrus growers seek rootstocks that induce smaller trees to allow high-density plantings, in order to facilitate management and harvesting, increase yield, and provide a high fruit quality (Carvalho et al., 2019a).In addition, it is important that the rootstocks chosen for each scion provide a high production efficiency, as well as resistance to the main diseases and abiotic factors that affect citrus orchards (Castle et al., 2010;Domingues et al., 2018).
'Swingle' citrumelo is the main rootstock used in Florida, USA, and one of the most important in Brazil (FDACS, 2018;Carvalho et al., 2019b).This rootstock induces a high orange production and high contents of total soluble solids, besides tolerance to citrus blight, sudden death, tristeza virus, and Phytophthora spp.gummosis (Castle et al., 2010).Considering the importance of this trifoliate hybrid, other rootstocks, designated as F.80 series, obtained from a similar crossing ('Duncan' citrumelo × P. trifoliata), have most of the features of 'Swingle' citrumelo and induce trees with a low vigor and small size (Pompeu Junior & Blumer, 2011).
The objective of this work was to evaluate tree size, production, and fruit quality of 'Valência' sweet orange grafted on various trifoliate orange rootstocks, in order to select genotypes with a high performance.

Materials and Methods
The experiment was conducted at the Experimental Research Station of Cocamar Cooperative in the municipality of Guairaça, located in the northwestern region of the state of Paraná, Brazil (22°56'30"S, 52°43'48"W, at 470 m above sea level), during consecutive harvest seasons.The climate of the region is Cfa, according to Köppen's classification, subtropical with hot summers, infrequent frosts, and concentration of the rainiest period in the summer months, but without a well-defined dry season.The average annual rainfall ranges from 1,400 to 1,600 mm, the average temperature from 22.1 to 23.0°C, and relative humidity from 65 to 70% (Nitsche et al., 2019).The soil is medium textured and composed of sand (or sandstone) on horizon A and clay on horizon B, being classified as a Argissolo Vermelho-Amarelo (Santos et al., 2018), i.e., an Oxisol.Additional information about the weather conditions during the trial, including temperature, rainfall, and water balance are available at the website of IDR-Paraná (2021).
Field-ready and grafted 'Valência' sweet orange trees of clone IAC (C.sinensis) (Pratinhas's Citrus Nursery, Paranavaí, PR, Brazil), were planted in April 2013 in a 6.0×2.5 m spacing, at 667 trees per hectare, under rainfed conditions.The randomized complete block design was used as a statistical model, with four replicates and four trees per plot.
Tree growth development, production, and orange juice chemical and industrial properties were assessed following the procedures described by Domingues et al. (2021b) and Association of Official Analytical Chemists (AOAC) (Latimer, 2019).Tree growth development (height, diameter, and canopy volume) was evaluated in mid-fall, from the 2016 to 2018 harvest seasons.Production per tree was determined from 2016 to 2019, and productive efficiency was obtained from 2016 to 2018.The chemical composition (soluble solids, titratable acidity, and maturation index or ratio) and industrial properties (juice yield, technological index, and industrial index) of the orange juices were evaluated from the 2017 to 2019 harvest seasons by the Integrada Cooperative in the municipality of Uraí, in the state of Paraná, Brazil, according to AOAC (Latimer, 2019).
Data were subjected to the analysis of variance, and means were grouped by the Scott-Knott test, at 5% probability, using the Sisvar software (Ferreira, 2011).In addition, the principal component analysis (PCA) was performed to explain the interrelationships between the rootstocks regarding the presented variables (Costa et al., 2020), which were subsequently grouped using hierarchical cluster classification (HCC).The R software was used to perform the multivariate analyses (PCA and HCC) with the aid of the FactorMineR package (R Core Team, 2019).
A quick reference chart was proposed based on Castle et al. (1993) and using: cluster analysis of production per tree; productive efficiency; technological index; industrial yield in the 2016, 2017, and 2018 harvest seasons; and canopy volume in the 2018 harvest season.
According to Costa et al. (2020), citrandarin and citrumelo rootstocks, evaluated in several trials and different environments, presented an interesting growth development and were among the most promising new hybrid rootstocks for sweet orange trees.In the present work, some genotypes showed a competitive performance in relation to 'Rangpur' lime under rainfed cultivation in subtropical climate (Cfa type) conditions.
When choosing scion/rootstock combinations, those that tend to take longer for the tree canopies to begin to overlap and form a hedgerow are preferred, because they would reduce the need for mechanical hedge pruning in the first years of tree growth development.Canopy overlap between rows hinders management practices and harvesting operations such as the transit of pickers, placement of harvest boxes for the temporary storage of fruit, and entry of agricultural machinery.
Therefore, frequent pruning is required to avoid overlap in high-density planting, and scion/rootstock combinations with shorter and narrower canopies may reduce the need for pruning in the early years of orchard establishment.For instance, in the present trial, due to the low canopy volume of 'Valência' sweet orange trees induced by F.80-18 and F.80-5 citrumelos and 'Flying Dragon' trifoliate orange, hedge pruning may only be necessary after the fifth year of orchard planting, considering the used spacing.Therefore, to establish 'Valência' orange orchards with small and low-vigor trees, F.80-18 and F.80-5 citrumelos can be considered alternative rootstocks, with dwarfing characteristics similar to those of 'Flying Dragon' trifoliate orange.It should be noted that, during the assessment period, none of the selected rootstocks presented graft incompatibility with 'Valência' sweet orange.
The highest means for accumulated production per tree along the seasons (Table 2) were obtained on 'US-852', 'US-801', IPEACS-256, and IPEACS-264 citrandarins, as well as on 'Rangpur' lime × sour orange and 'Rangpur' lime, while the lowest ones were observed on 'Frying Dragon' trifoliate orange.However, in 2018, production efficiency was the highest for trees grafted on F.80-18, followed by those on F.80-3 and F.80-5 citrumelos, with values that were even higher than those on 'Rangpur' lime and 'Swingle' citrumelo (Table 3).Production efficiency was, on average, ≈200 and 300% higher on F.80-18, F.80-3, and F.80-5 citrumelos, compared with 'Rangpur' lime and 'Swingle' citrumelo, respectively.Under different environmental conditions, those rootstocks also induced a higher production efficiency for 'Valência' sweet orange than 'Rangpur' lime and 'Swingle' citrumelo, but in a lower proportion (Pompeu Júnior & Blumer, 2011).Therefore, the evaluation of the performance of rootstocks in different edaphoclimatic conditions is essential to identify those that gather the most desirable features for each site.
Under the environmental conditions of this trial, high-density plantings with rootstocks with a low vigor and high production efficiency, such as F.80-18, F.80-5, and F.80-3 citrumelos, could make the production system of new 'Valência' sweet orange orchards more feasible, with greater productivity earnings.This is important since high-density planting is one of the strategies used to mitigate HLB -the main disease that currently affects orange orchards worldwide -, allowing the removal of symptomatic trees (Vashisth et al., 2020).
In the last season assessed, when 'Valência' orange trees were considered mature, the highest means for the orange juice maturation index or ratio, which is directly influenced by the soluble solids and titratable acidity contents of the orange juices, were observed on IPEACS-256 citrandarin, F80-18, F.80-3, F.80-5, F.80-6, F.80-8, and 'Swingle' citrumelos, 'US-802' pummelo hybrid, and 'Flying Dragon' trifoliate orange; the obtained values were higher than those on 'Rangpur' lime (Table 4).In addition, the rootstocks had no influence on juice yield, but the industrial yield Table 2. Production per tree and accumulated production per tree of 'Valência' sweet orange (Citrus sinensis) grafted on 20 rootstocks in the 2016, 2017, 2018, and 2019 harvest seasons (1) .
In other words, when 'Valência' sweet orange trees were grafted on those rootstock genotypes, compared with on 'Rangpur' lime, there was a lower demand (≈10% less) for orange boxes to produce a ton of frozen concentrated orange juice, implicating a better processing efficiency.
For the principal component analysis, the industrial yield and technological index had greater contributions, showing a negative correlation between both of them (Figure 1).Yield (production per tree) was positively correlated with canopy volume, indicating that vigorous trees had higher yields, but was negatively correlated with productive efficiency, as also reported by Fadel et al. (2018) and Rodrigues et al. (2019).
It was possible to identify four groups of rootstocks.The first two groups induced better quality 'Valência' orange fruits and were composed of: 'US-801', 'US-812', and 'Swingle' citrumelos, 'US-802' pummelo hybrid, trifoliate orange, and 'Flying Dragon' trifoliate orange.The third group was composed of 'US-852' citrandarin, F.807 and 'W2' citrumelos, 'Rangpur' lime, 'Murcott' tangor × trifoliate-9, and 'Rangpur' lime × sour orange, which induced a higher production, but were vigorous and had an inferior performance regarding fruit quality.The last group was composed of F.80-18 citrumelo, completely isolated from the other rootstock genotypes, showing a better production efficiency.The multivariate analysis provided more comprehensive information about the performance of the assessed rootstocks, identifying homogenous groups of genotypes, considering the universe of all characteristics that were significant by the univariate analysis of variance, as also described by Costa et al. (2020Costa et al. ( , 2021)).
Based on the findings of the present study, a quick reference chart was proposed, accounting for the influence of each rootstock on the main characteristics of 'Valência' sweet orange trees, aiming for a practical and objective identification of the best rootstock alternatives (Table 5).In particular, for high-density plantings, F.80-5 citrumelo is an interesting option because it induces dwarf trees, a high productive efficiency, and a high industrial quality of fruits, followed by F.80-3 citrumelo and IPEACS-239 citrandarin.Moreover, in scenarios where the establishment of low-density plantings is preferred, 'US-801', 'US-812', and IPEACS-256 citrandarins are better options than 'Rangpur' lime and 'Swingle' citrumelo because 'Valência' sweet orange trees grafted on those genotypes have a lower tree size and a better yield and industrial quality.  (1. Rootstock (2)  Productive efficiency (kg m   (1) .2. Rootstocks of F.80-5 and F.80-3 citrumelos and IPEACS-239 citrandarin are also adequate options to obtain orange juices with good industrial properties.

Table 4 .
Chemical and industrial properties of juices of 'Valência' sweet orange (Citrus sinensis) trees grafted on 20 rootstocks in the 2017, 2018, and 2019 harvest seasons Means followed by equal letters do not differ by the Scott-Knott test, at 5% probability.Trees were planted in April, 2013.CV, coefficient of variation. (3)Number of boxes of 40.8 kg per ton of frozen concentrated orange juice. (4)Kilogram of SS per box of 40.8 kg.CV, coefficient of variation.* and **Significant at 5 and 1% probability, respectively.