Effect of seventeen rootstocks on young ‘Valência’ sweet orange performance in western Santa Catarina, Brazil

Recommendations of rootstocks for ‘Valência’ sweet orange in southern Brazil have been partially based on empirical observations or information. Therefore, this study compared young ‘Valência’ trees budded on different rootstocks in western Santa Catarina, Brazil. Seventeen rootstocks were tested for growth (tree height and canopy projection area and volume) and fruit production (per tree, hectare, m2 and m3 of canopy). The experiment comprised four replicates of three trees and was carried during six years under standard management, in a Cfa (humid subtropical) climate. In general, larger trees produced more fruit. However, ‘San Diego’ citrandarin, ‘Swingle’ citrumelo, ‘Cravo’ rangpur lime (among vigorous rootstocks), ‘Cravo’ x ‘Sunki’ EEI hybrid, ‘Fepagro C37 Dorneles’ citrange, ‘Rubidoux’ trifoliate orange (among medium vigorous rootstocks), ‘HFD25 EEI’ hybrid and ‘Flying Dragon’ trifoliate orange (among less vigorous rootstocks) stood out in cumulated production by area or volume of the canopy. ‘San Diego’ also stood out in production per tree. Cultivars ‘Cravo’ x ‘Sunki’ EEI, ‘San Diego’, and ‘HFD25 EEI’ should be further evaluated regarding disease susceptibility and fruit quality for possible release.


INTRODUCTION
Environmental stress in the form of droughts, floods, frosts, high temperature, pests and diseases are factors that can negatively affect citrus trees growth. In order to overcome these challenges, growers can combine scion and rootstock varieties through grafting (Ribeiro, Espinoza-Núñez, Pompeu, Mourão Filho and Machado, 2014). When choosing a rootstock, growers must take into account a list of attributes. Castle (2010) suggests twenty-four attributes concerning yield and reaction to biotic and abiotic soil characteristics, among others.
In subtropical southern Brazil, 'Valência' sweet orange [Citrus sinensis (L.) Osbeck] is an important scion cultivar, especially due to its late harvest. However, sweet oranges are susceptible to soil diseases that damage roots and the trunk collar, such as Phytophthora sp. root rot. This leads growers to graft 'Valência' onto Phytophthora tolerant rootstocks if the orchard's soil is finetextured. In addition to disease tolerance, tree vigor and yield, cold hardiness, and fruit quality are important traits to take into account to choose a rootstock. Trifoliate orange [Poncirus trifoliata (L.) Raf.] and its hybrids (citranges and citrumelos) have been recommended by citrus experts and governmental agencies. However, especially in Santa Catarina State, citrus nurserymen prefer fast growing rootstocks, which shorten nursery tree growing time, such as 'Cravo' (Rangpur) lime (Citrus limonia Osbeck), a Phytophthora sensible genotype, and 'Swingle' citrumelo (Citrus paradisi Macfad x P. trifoliata).
In general, recommendations of rootstocks for 'Valência' orange cultivation in the states of Rio Grande do Sul and Santa Catarina are partially based on empirical observations or data from other scion cultivars, since scientific works regarding rootstocks for 'Valência' are scarce. Some studies have already been carried in medium-textured soils of southern Rio Grande do Sul (Koller, Marodin, Schwarz and Nienow, 1988;Koller, Schwarz and Panzenhagen, 1999;Kirinus et al., 2019), therefore, this study compared the agronomic performance of young 'Valência' sweet orange grafted on 17 rootstocks, in western Santa Catarina, Brazil.

MATERIAL AND METHODS
The experiment was carried out in the municipality of Águas de Chapecó, Santa Catarina State, Brazil, in the Rio Uruguay valley region, at 330 m above sea level, and in a humid subtropical climate (Koppen -Cfa). The soil (Nitisol, 30-40 % clay) was previously corrected for pH, P and K levels. Standard 'Valência' sweet orange nursery plants were transplanted in May 2013, following the experimental treatments (rootstocks) described below and spaced 7 m x 3 m.
The experiment was designed in four completely random blocks with three-tree plots. Seventeen rootstocks genotypes were tested: Trees were grown following standard procedures for orange growing in Santa Catarina State. Fertilizations were performed considering the expected average production (Rockenbach and Koller, 2013), keeping the same doses across the treatments. In August 2013, all trees were pruned at 50 cm from the ground, selecting later three to four new shoots for canopy formation. No further pruning was made to avoid limiting expression of rootstock vigor. Fungicides were regularly sprayed during flowering and fruitlet times.
Tree height (H) and canopy diameter (D) (average of transversal and longitudinal to the row) were measured in May 2019. This basic data was used to calculate additional output variables, namely canopy volume (V) (V = 2/3 π×H×D) and projection area (PA) (PA = π×(D/2)^2), canopy shape index (H/D ratio), cumulated fruit production per tree (ignoring dead ones) for 3-6 years (P), cumulated fruit production per hectare considering zero the production of dead trees, productive efficiency index by volume (P/V ratio), and area (P/ PA ratio).
The output variables were submitted to the Box-Cox test to verify data suitability (and transform it if necessary) for further analysis of variance (ANOVA). The rootstock means were compared by the Scott-Knott grouping test (α=0.05). The relation between variables was analyzed through the Principal Component Analysis (PCA). All analyses used "R" software provided with "ScottKnott", "MASS" and "Agricolae" packages (R Core Team, 2018).

RESULTS AND DISCUSSION
All eight variables of response were significantly affected by the rootstocks (Table 1). 'Valência' trees grew less when budded on 'Flying Dragon' and 'HFD 25 EEI' (1.51 m and 1.72 m high, respectively) and grew more on a group of seven rootstocks whose heights were statistically the same (3.13 m to 3.56 m) (Figure 1a). Other eight treatments had an intermediate height (2.46 m to 2.89 m). 'Flying Dragon' and 'HFD 25 EEI' showed a significantly smaller shape index too, lower than 0.89, while the other 15 rootstocks ranged from 0.97 to 1.15 without differing significantly from each other (Figure 1b). to follow the tree height. This relation is shown in Figure 2, where the arrows corresponding to the three variables above point to approximately the same direction. In the Scott-Knott classification using the average of canopy volume and projection area, the only difference from height classification was 'Cravo', which changed from the mediumheight class to the high-volume and high-area classes (Figures 1c-d).

Canopy volumes and projection areas tended
Valencia on 'San Diego' citrandarin showed the highest cumulated production per tree (212.13 kg) (Figure 1e). Seven rootstocks comprised a second group, producing 116.39 kg to 176.61 kg, while other seven produced 74.59 kg to 110.09 kg. 'Flying Dragon' and 'HFD 25 EEI' produced the smallest amount of fruit: 40.83 and ) Osbeck] tree height, canopy shape index, volume¹ (V), and projection area (PA), production per tree¹ (P) and per hectare¹, P/PA and P/V¹ ratios, cumulated until the fourth harvest season in Águas de Chapecó, Brazil (2013-2019. Bars filled with the same pattern represent no significant differences (Scott-Knott test, α=0.05). ¹Variable was transformed 54.72 kg tree -1 , respectively. Since all treatments had the same tree density (476.19 trees ha -1 ), yield per hectare was proportional to tree production, except for 'Cravo', which had three dead trees across two blocks, attributed to Phytophthora sp. rot. Therefore, its production per hectare was proportionally smaller. On the other hand, 'Cravo' showed production per tree and per hectare similar to 'Swingle', 'Sunki' x 'Benecke', 'BRS CNPMF Tropical' and 'Carrizo'. The significant differences between rootstocks regarding tree production were not the same as those for production per hectare (Figure 1e The two principal components in the PCA explained 96 % of the variance in the data. They revealed a strong positive relation between growth variables (height, V, and PA) and a strong negative relation between the shape index and P/V ratio, meaning that plants that were more prostrated tended to be more efficient. 'Flying Dragon', for instance, is known by its higher production efficiency per unit of canopy volume induced to the scion (Cantuarias-Avilés, Mourão Filho, Stuchi, Silva and Espinoza-Nuñez, 2011). 'Cravo' is still the main rootstock used in Santa Catarina State, although researchers do not recommend its use due to its sensibility to Phytophthora sp. root rot and frosts and induction of poor scion fruit quality. Trifoliate orange and its hybrids, such as 'Swingle' citrumelo, have been promoted as better genotypes to produce good quality fruits. 'Cravo' and 'Swingle' showed good growth and productive performance in this experiment with 'Valência', although some 'Cravo' trees have died. 'Swingle' has been the main option for growers, since it is well adapted and forms vigorous, good quality fruit producing trees.
The results presented here highlight 'San Diego' citrandarin as a potential rootstock for 'Valência' in southern Brazil and probably in neighboring countries, since it showed the highest yield in the first four harvest seasons (Figures 1e-1f). In Pelotas-RS, 'San Diego' induced 'Valência' to produce fruit with better color and high polyphenol content, with good yield and vigor (Kirinus et al., 2019). In São Paulo State, 'San Diego' induced 100 % survival rate, low alternate bearing, high drought tolerance and yield (Costa, 2019).
Among the rootstocks of intermediate tree size, hybrid 'Cravo' x 'Sunki' EEI (unreleased genotype), combined to 'Valência', formed a highly efficient canopy, producing higher fruit mass per m³ than the rootstocks with similar tree size. However, as 'Cravo' and 'Sunki' genitors have some major weaknesses, the hybrid may have inherited them, especially Phytophthora sensibility (Medina Filho et al, 2003). These traits should be considered in future works before the release of the hybrid as a rootstock cultivar.
Hybrid 'HFD 25 EEI' as rootstock for 'Valencia' performed similarly to 'Flying Dragon', the main dwarfing citrus rootstock. Trees budded on these hybrids were smaller and more prostrated (canopy width larger than height) than the others, with good production per m³. The PCA revealed ( Figure  2) that the higher the shape index (more vertical canopies), the lower the P/V ratio. It possibly happens because fruiting limits plant growth (Lliso, Forner and Talón, 2004) and fruit weight causes branches to become decumbent, decreasing the shape index.
The difference between trees of the same scion variety budded on different rootstocks in the same environment are caused, at least in Figure 2. Graphical representation of the two main components PC1 and PC2 (Principal Component Analysis) explaining variance in the observed data for 'Valência' sweet orange [Citrus sinensis (L.) Osbeck] budded on 17 rootstocks. PC1 and PC2 altogether explain 96 % of data variance. The variables analyzed are: tree height, shape index, canopy volume (V) and projection area (PA), tree production (P), P/V and P/PA ratios part, by the genetic background of the rootstock, which expresses different phenotypical traits. For example, 'Cravo' roots grow deeper than trifoliate orange roots do (Magalhães Filho, Amaral, Machado, Medina and Machado, 2008). Therefore, trees on 'Cravo' can explore a larger soil volume, where it absorbs water and nutrients. Moreover, different root systems vary in growth regulators synthesis, transportation, and distribution resulting in different scion phenotype (Liu, Li, Liu, Yao and Chen, 2017). In addition, RNA-transcripts from rootstock genes can migrate to scion and viceversa, and the rootstock can modify the expression of some scion genes (Liu et al., 2018). Some rootstocks have the capacity to change assimilates distribution pattern across the organs as a dwarfing mechanism, that is, fruit load reduces tree growth, following competition for assimilates (Lliso et al., 2004).
Some combinations of rootstocks and scions are not physiologically and biochemically compatible, resulting in bad tissue union and tree decline (Pompeu and Blumer, 2014). This phenomenon could explain bad performance of combinations; however, incompatible combinations with 'Valência' scion seem to be unusual, as Costa (2019) found no evidence of incompatibility among 46 rootstocks with 'Valência'.
Considering that rootstocks provide different growth to scions, comparing ratios P/V or P/PA provides a better understanding of the rootstock value, since smaller trees can be grown in higher tree density. Therefore, it is better to compare rootstocks within each growth group (canopy volume or area) (Figures 1c-d). Higher efficiency stands out for 'San Diego', 'Swingle', 'Cravo', (big-tree group), 'Cravo' x 'Sunki' EEI, 'Fepagro C37 Dorneles', 'Rubidoux' (medium-tree group), 'HFD25 EEI', and 'Flying Dragon' (small-tree group). However, Koller et al. (1988) reported an increase in production efficiency by m² of 'Valência'/P. trifoliata from the fifth harvest onward and the opposite for 'Valência'/'Cravo', which means that the results in the present study may not be true for an adult orchard.
Other researchers have studied the same rootstocks in other environments. In Florida (USA), 'Valência' on 'Swingle', and trifoliate orange had similar heights; however, lower than on 'Carrizo' citrange (Castle, Baldwin and Muraro, 2010). In northwestern Paraná State, trees on 'Cravo' produced less than on 'Sunki', without differences neither in canopy height nor in yield efficiency (Auler, Fiori-Tutida and Tazima, 2008). In northern São Paulo State, 'San Diego' did not differ from 'Cravo' clones and 'Sunki Tropical' neither in canopy volume nor in tree production or production efficiency induced to 'Valência'. These findings are not in agreement with this work.
Different performances of a same scion/rootstock combination across sites could be attributed to differences in environmental conditions, such as soil biotic and abiotic factors, weather, orchard management, and their interactions. In addition, slight genetic differences may occur between accessions of the same variety (same name) (Snoussi et al., 2012).