Performance of 'Jaffa' sweet orange on different rootstocks for orchards in the Brazilian Northeast

Abstract: The objective of this work was to evaluate the effect of nine rootstocks on 'Jaffa' sweet orange (Citrus sinensis) for its vegetative, productive, and qualitative traits, and for its susceptibility to three pest mites (Eutetranychus banksi, Tetranychus mexicanus and Phyllocoptruta oleivora). The following parameters were evaluated: vegetative growth, yield, physiochemical characteristics of fruit, and density of pest mites. 'Sunki Tropical' mandarin conferred high cumulative yields whereas 'Orlando' tangelo and 'Rugoso Vermelho' lemon imparted bigger fruit with low brix, and less acidity to 'Jaffa'. Indio and 'Riverside' citrandarins induced more acid fruit of smaller sizes, as well as low-cumulated yields and canopy volumes. Also, the 'Cravo Santa Cruz' lime, 'San Diego' citrandarin, and the hybrids HTR-051 and LVK x LCR-010 conferred lower-cumulated yields to 'Jaffa'. The rootstocks did not influence the population levels of the evaluated mites. 'Sunki Tropical' mandarin, 'Cravo Santa Cruz' lime and 'Rugoso Vermelho' lemon stand up as excellent rootstock options for 'Jaffa' sweet orange in the Northeastern Brazil.


Introduction
Brazil ranks first in the production of sweet orange in the world, and Brazilian Southeast and Northeast are the main producing regions (IBGE, 2016). In Northeastern Brazil, the majority of orchards cultivate 'Pera' sweet orange [Citrus sinensis (L.) Osbeck] grafted on 'Rangpur' lime (C. limonia Osbeck) because of fruit with good quality and drought tolerance imparted from the rootstock (Almeida & Passos, 2011). However, such a narrow genetic diversity could lead to losses caused by biotic and abiotic stresses, threatening the whole citrus chain in this region (Carvalho et al., 2019). Accordingly, Embrapa (Brazilian Agricultural Research Corporation) begun field trials to select new scion-rootstock combinations for rainfed citrus groves of the Brazilian Northeast, which resulted in the indication of some cultivars (Carvalho et al., 2016a(Carvalho et al., , 2016b(Carvalho et al., , 2019. 'Jaffa' orange originates from Israel, where it is the most cultivated scion cultivar (Ishfaq et al., 1999). It is characterized by medium-sized plants, with open branches, satisfactory yields, tendency to production alternation, precocious maturation, and fruit suited for both in natura consumption and juice production (Bacar et al., 2017). Some pests, among which the Texas citrus mite Eutetranychus banksi and the tetranychid Tetranychus mexicanus (Acari: Tetranychidae), as well as the rust mite Phyllocoptruta oleivora (Acari: Eriophyidae) attack citrus orchards in Northeastern Brazil (Teodoro et al., 2014;Silva et al., 2016). While E. banksi and T. mexicanus feed upon leaves of citrus trees, P. oleivora is a fruit-attacking pest that leads to reduced yield and aesthetic damage to the fruit that is similar to rust (Teodoro et al., 2014). Therefore, assessing the influence of rootstocks on the susceptibility of scions to pests is highly recommended, as cultivars show distinct genetic characteristics (Vendramim & Guzzo, 2012;Silva et al., 2016). For instance, densities of P. oleivora and T. mexicanus on 'Pera CNPMF D-6' and 'Valencia Tuxpan' sweet oranges were influenced by some rootstocks (Silva et al., 2016).
The objective of this work was to evaluate nine rootstocks on the vegetative, productive, and qualitative traits of 'Jaffa' sweet orange, as well as on the susceptibility of this cultivar to three pest mites.
Plant height (PH), canopy volume (CV), rootstock/ scion girth ratio (R/S), and yield efficiency (YE) were measured in 2016, as follows: PH (m) was recorded by using a ruler from the base of the trunk close to the soil up to the top of the plant; CV (m 3 ) was estimated using the expression CV = (π/6) × PH × DR × DP, in which DR and DP are canopy diameters (m) along the row, and in the perpendicular direction to the row, respectively (Cantuarias-Avilés et al., 2012); yields and cumulative yields (CY, in kg ha -1 ) were evaluated over 6 years (2011 to 2016), while yield efficiency (YE, in Pesq. agropec. bras., Brasília, v.55, e01665, 2020 DOI: 10.1590/S1678-3921.pab2020.v55.01665 kg m -3 ) was estimated by the quotient between the per plant fruit production and CV. Fruit quality was assessed over 2 years (2014 and 2015) by collecting six fruits per plant (Carvalho et al., 2016a). Briefly, height, rind thickness (mm) and diameter of fruits were measured with a digital caliper (Carvalho et al., 2016a). Juice content was estimated with the equation JC = [(juice mass / sample mass) × 100] (Carvalho et al., 2019). Juice mass was obtained by the difference between sample weight and bagasse mass (that is, seeds, peels, and fragments retained in the sieve) (Carvalho et al., 2016a). Total soluble solids (TSS in ºBrix) were recorded with a digital refractometer with values corrected to 20°C (Carvalho et al., 2016a). Titratable acidity (TA, g citric acid 100 mL -1 juice) was determined with a semiautomatic burette with 0.1 mol L -1 NaOH (titrant) and a phenolphthalein indicator. The ratio was determined as the quotient between TSS and TA (Carvalho et al., 2016a). The content of vitamin C (mg 100 mL -1 of juice) was calculated using the oxidation-reduction volumetric technique, using 0.002 mol L -1 potassium iodate (KIO 3 ) and 1% starch indicator solutions (Carvalho et al., 2016a). Industrial yield was determined with the equation IY = 660/TI, in which TI (technological index) was obtained as juice content x TSS × 40.8/10.000, which is equivalent to the amount of TSS in the juice (kg) in a standard 40.8 kg industrial box of citrus (Carvalho et al., 2016a). Data were subjected to the analysis of variance, followed by Scott-Knott's tests, at 5% probability.
Multivariate analyses were also performed using XLSTAT (a statistical software for Excel) to identify homogenous groups of rootstocks, considering the universe of all variables that were significant by univariate analysis of variance. At first, a principal component analysis (PCA) was conducted to reduce the dataset into a few synthetic and uncorrelated variables, that is, the first principal components (Carvalho et al., 2019). Then, the rootstocks were grouped by agglomerative hierarchical clustering (AHC) applied on the PCA scores, using the Euclidean distance as a measure of dissimilarity and Ward's minimumvariance method for linkage to identify the clusters (Carvalho et al., 2019). The automatic truncation option was used for cluster splitting. This approach creates homogenous groups based on the largest decrease in the Shannon's entropy between a node and the next one (Carvalho et al., 2019). The resulting clusters were interpreted by means of the PCA results and put into perspective with the results of univariate analyses of variance.
The effect of rootstocks on the population levels of three regional pests of citrus, namely E. banksi, T. mexicanus, and P. oleivora was assessed. The number of adults of these mites were monthly recorded from June 2011 to February 2013 (counts of some months were excluded from the analyses as they were null to all rootstocks). In short, E. banksi and T. mexicanus were recorded in four randomly chosen leaves from two plants per plot, totaling 32 leaves per rootstock in each monthly evaluation. For P. oleivora, two randomly-chosen fruit taken from two plants per plot were evaluated, totaling 16 fruit per rootstock in each evaluation, and mites were counted in 1 cm 2 area. Repeated measures of the analyses of variation, followed by the post hoc Fisher's LSD tests, were carried out to evaluate the effect of rootstocks on the population levels of the pest mites, removing the variance explained by time, as evaluations were conducted monthly. Data were √(x) + 0.5 transformed. in order to adjust them for normal distribution (Silva et al., 2016).

Results and discussion
Rootstocks did not influence the height of 'Jaffa' sweet orange plants, in contrast with the highestcanopy volumes conferred by 'Sunki Tropical' mandarin, 'Rugoso Vermelho' lemon and the hybrid LVK x LCR-010 (Table 1). However, rootstocks did not affect the yield efficiency which considers the yield per volume of canopy.
'Cravo Santa Cruz' lime, 'Riverside' and 'Indio' citrandarins, 'Sunki Tropical' mandarin, and LVK x  LCR-010 imparted higher TI values to 'Jaffa' than the remaining rootstocks. The average TI for all rootstocks was 2.25 kg TSS per box, which was higher than that obtained by Bacar et al. (2017). Variables that were significant by the univariate analyses of variance and used for the multivariate analysis were: canopy volume (CV) and cumulative yield (CY, 2011(CY, -2017 for vegetative and productive performance and fruit diameter (FD); fruit height (FH); juice content (JC); titratable acidity (TA); total soluble solids (TSS); TSS/box; industrial yield (IY); and TSS/ TA ratio (ratio) for fruit quality. About 75% of the total variability could be explained by the first two PCA axes (Figure 1), and the square cosine of the variables   showed that fruit diameter (FD), fruit height (FH), TSS, TSS/box, and IY were the most explanatory variables of PC1, whereas JC, TA, CV, and CY were the most explanatory variables of PC2. A significant positive correlation was observed between cumulative yields and canopy volumes, indicating that the greater the canopy, the higher the yield, especially because rootstocks did not significantly affect the yield efficiency of 'Jaffa'. Working with three different orange scions, Carvalho et al. (2019) identified that while the trifoliate rootstock HTR-051 induced dwarfism in 'Sincora', 'Valencia Tuxpan', and 'Pineapple', such dwarfism was associated with an increased yield efficiency (that is, mass of fruit per volume of canopy) only for 'Sincora', concluding that the adequacy to high-density plantings depends upon a scion/rootstock combination. In the present work, even though some rootstocks induced plants of small volumes, such dwarfism was not followed by an increased yield efficiency, which indicates that 'Jaffa' is not a sweet orange scion adapted to high-density plantings.
In general, fruit size [small diameter (FD), and height (FH)] were negatively correlated with TSS, TSS/box, and TA, indicating that bigger fruit normally showed low-Brix values and less acidity. Even though the correlation between fruit size (FD and FH) and CY or JC could not be detected in the present study, the rootstocks that induced higher cumulative yields were also those that showed lower-juice contents in the fruit. The AHC grouped the rootstocks into four groups. The first group comprised only 'Sunki Tropical' mandarin, and was characterized by high cumulative yields. The second group included 'Orlando' tangelo and 'Rugoso Vermelho' lemon, and showed bigger fruit with low-Brix, low-SST/box, and less acid values. On the opposite, the third group, that included the 'Indio' and 'Riverside' citrandarins, tended to show more acid fruit, as well as low-cumulated yields and canopy volumes. Finally, the fourth group, that contained 'Cravo Santa Cruz' lime, 'San Diego' citrandarin, as well as the hybrids HTR-051 and LVK x LCR-010, shows mostly lower cumulative yields than the group 1, and intermediate values for the other evaluated variables. This result indicates that while 'Sunki Tropical' mandarin is the best rootstock for promoting vegetative development and yield, the group composed by 'Cravo Santa Cruz' lime, 'San Diego' citrandarin, and the trifoliate hybrids HTR-051 and LVK x LCR-010 induced fruit of better quality.
2. The evaluated rootstocks do not influence the population levels of the pest mites E. banksi, T. mexicanus, and P. oleivora in 'Jaffa' sweet orange.