TOP GRAFTING TO ACCELERATE SELF-POLLINATION IN Eucalyptus BREEDING

The top grafting technique can make early fl owering possible and consequently accelerate self-pollination in Eucalyptus breeding, reducing the period of each generation. This study aims to establish a methodology to top graft by applying the growth regulator paclobutrazol in self-pollinated Eucalyptus genotypes to induce early fl owering and accelerate inbred line strategies. A total of 448 top grafts of seven genotypes (three Eucalyptus urophylla × Eucalyptus grandis hybrids, one Eucalyptus urophylla, and three Eucalyptus grandis) were performed in two periods of the year: July and October. The top grafting samples were evaluated concerning fl ower induction and graft development at three-month intervals. A t-test was performed with a 5% signifi cance level for type I error to compare the relevance of paclobutrazol application. A fi xed model was also used to analyze the signifi cance of the treatments. The fastest blooming occurred after three months of applying the graft methodology. After two years, the top graftings performed in October presented higher fl ower bud and fruit production. The top grafting aff ected the induction of self-pollinated Eucalyptus fl owers with enough fl ower buds produced to follow the next cycle of self-pollination in some genotypes. The eff ectiveness of selfpollinated top graftings varied with the genotype used as the scion. The paclobutrazol improved the fl owering of the top grafting samples. The methodology established in this work allows accelerating self-pollination strategies in the globally important industrial crop Eucalyptus.


INTRODUCTION
Self-pollination strategies have high potential to increase genetic gains related to wood quality when individuals within the lines are generated, maximizing heterosis for yield traits by crossing line individuals (Cobb et al., 2019;Saxena et al., 2021). Inbred line strategies are commonly used in agronomic crops (Wang et al., 2012;Gasim et al., 2015;Avdikos et al., 2021). This strategy aims to increase the homogeneity of seminal plantings by crossing individuals with elevated levels of homozygosity (Maia, 2010;Salvador et al., 2021). Applying these procedures in Eucalyptus can increase the additive genetic variance among inbred lines after self-pollination cycles and reduce the variance within lines . The inbred lines could allow for selecting individuals with dominant homozygous alleles for the trait of interest, contrasting with individuals from other inbred lines. The maximum expression of additive eff ects within lines and heterosis in the crossline genotypes can be explored due to the high additive genetic variance in crossing inbred lines (Reddy et al., 2015;Santos et al., 2016). Therefore, self-pollination strategies can generate considerable gains for the forest industry by generating elite genotypes.
The potential of self-pollination strategies in forest breeding programs is high, but the up to twoto-seven-year time for the trees to mature and fl ower makes generating inbred lines impractical (Jones et al., 2011;Klocko et al., 2016). The performance of consecutive cycles using traditional breeding is not feasible because obtaining homozygous individuals takes up to six or seven cycles of self-pollination and selection (Ramalho and Araújo, 2011). Selfpollination is common in agronomic crops, but it has not been utilized in the forest sector due to the long generation period. New techniques such as top grafting with early fl ower induction have the potential to accelerate self-pollination cycles and increase their genetic gain Wong and Bernardo, 2008;Grattapaglia et al., 2018;Castro et al., 2021).
The top grafting technique allows preserving individuals with relevant botanical, morphological, or ecological characteristics by propagating them effi ciently with earlier bloom induction (Almqvist, 2013a;Tabacu et al., 2020). Top grafting uses the canopy of physiologically mature and reproductive adult trees, optimizing an available rootstock (Gaspar et al., 2017). This technique has been used in Pinus and fruit growing trees, such as kiwi (Liang et al., 2011;Almqvist, 2013b;Gaspar et al., 2017) and is promising for improving the Eucalyptus genetic material in breeding programs. Eucalyptus species are long cycle allogamous. The best individuals of the breeding population are crossed, and their best progenies are cloned. However, the possibility of increasing the heterosis and obtaining signifi cant gains in traits of interest can be achieved using inbred lines. Therefore, there is a need to adapt this methodology for Eucalyptus trees and obtain the inbred lines in a shorter time. Selection of the best individuals in each cycle with the use of top grafting and molecular markers to accelerate the fl owering and the consecutive self-pollination cycles, until achieving the desired homozygosity rate, reduces the necessary time and adverse eff ects of inbreeding (Bison et al., 2004;Nickolas et al., 2019).

The potential of inbred lines developing in
Eucalyptus and the problems they present justify establishing a viable methodology to perform top graftings. Therefore, the aim of this study is to establish a methodology to top graft and test whether it can be combined with application of the growth regulator paclobutrazol (PBZ) in self-pollinated genotypes, aiming at induction of early fl owering and acceleration of inbred line strategies.

Environmental conditions and genetic material
The experiment was conducted in the hybridization orchards CENIBRA S.A. (latitude 18º 46′ 30″ S; longitude 42º 55′ 57″ W; altitude 744 m), Minas Gerais, Brazil. The climate is classifi ed as Cwa according to the Köppen and Geiger classifi cation with an annual rainfall of 1,497 mm and average temperature of 19.9 oC. Selected genotypes were self-pollinated using pollen isolation bags, and their seeds were planted in 2012 as the fi rst cycle of inbred progeny tests of the species Eucalyptus grandis W. Mill ex Maiden, E. urophylla S.T. Blake, and hybrids of E. urophylla × E. grandis. After the fi rst cycle of self-pollination, seven self-pollinated individuals were selected to be evaluated through top grafting for inbred line production. The choice of these genotypes was determined by their volume and wood density values measured from the inbred progeny test at 5 years of age. Twigs were collected in 2017 from the trees and used as scion in the top graftings.
Rootstocks with the same genetics (maternal parents of the genotypes from the fi rst inbred cycle) as the scion were used to make the compatibility between the graft and the rootstock feasible. The ages of these trees were between 18 to 32 years and each genotype was replicated twice in the company's orchard for the top graftings, totaling 14 rootstocks. These old trees were being underused, so using them as rootstock for this methodology was fi nancially expedient. The eff ect of the growth regulator (PBZ) in these replicas was evaluated on the survival and development of top graftings. The rootstocks were physiologically mature (blooming) with the potential of producing abundant fl owers and fruits due to its good crown formation, size, and phytosanitary status.

Experimental design
Two eff ects were evaluated to establish a viable methodology using the graft technique to accelerate self-pollinating generations of Eucalyptus: a) the application or absence of PBZ and b) the best period of the year for this procedure. The tested periods were six and three months before the conventional species fl owering season (January-March); that is, the grafts were performed in July (period 1) and October (period 2) of 2017, respectively. The lower physiological activity of plants in these periods due to the low temperatures tends to be benefi cial to the graft's survival (Gaspar et al., 2017;Perez-Luna 2020) as they only direct their compounds to induce sprouting and fl owering when the rainy season starts.
The experimental design was randomized blocks with eight replicates per rootstock, equally divided into the two crown sides with each face representing a time when the grafts were performed. Each replicate was established from the graft of four top graftings in a diff erent rootstock branch. In addition, one of the two replicates of the rootstocks of same genetics was selected to be used for the application of the regulator and another as a control without PBZ. The other treatments used in this study, including pruning, watering, and fertilization of the rootstocks, were the same with or without PBZ. Sixty-four top grafts were made per genetic material, with 32 in each of the rootstock replicates (with and without PBZ). Thus, a total of 448 top graftings [seven self-pollinated genotypes × four top grafts/rep × four reps/period/ side of the crown × two periods (six and three months before fl owering) × two treatments (with and without PBZ)] were made.

Top graftings of self-pollinated genotypes
The grafting technique used was the fork in full slit (top cleft) method, performed in the best branches of the rootstock canopy with some adaptations. The scion (twigs) were collected and transported to the adult rootstocks. Branches with twigs of the same scion circumference were selected from the rootstocks for grafting. The grafted material was coated with parafi lm. The connection between scion and rootstock was pressed with thread-seal tape. Three months after the graftings, the plants were evaluated.

Paclobutrazol application
After performing the top graftings, one of the rootstocks of each genotype was selected in the fi rst period (six months before fl owering) and PBZ was applied over its root area. This step started by measuring the circumference of the rootstock trunk base. The ratio used was 1 mL of Cultar 250 SC in concentrated suspension (250 g/L paclobutrazol composition), per centimeter of circumference. The regulator was diluted with fi ve liters of water per rootstock application. The PBZ infl uence on fl owering was evaluated at intervals of approximately three months after the grafts were made for each period (six and three months before fl owering). The data from each top grafting per genotype, in rootstocks with and without PBZ, were separated during the evaluations for analysis.

Data collection
The survival rate, fl owering, and canopy area were also measured in periodic evaluations at threemonth intervals. The survival rate was obtained each evaluation from the count of living top grafts over time. When fl owering was detected, the evaluations also consisted of counting the fl oral buds, fl owers, and fruits present. The top graftings' canopy area (m 2 ) was calculated by multiplying widths and lengths, both in cm, of all living grafts and dividing the result by 10,000 to convert the value to m 2 . The growth of the canopy over time was monitored with the area values.

Eq.1
The top graftings submitted to diff erent eff ects were compared using the collected data from eight evaluations (from September 2017 to August 2019). The vigor of the rootstocks was also evaluated. Specifi c recommendations regarding fertilization, orchard management, and graft maintenance and identifi cation were made to improve the material conditions over time.

Statistical analysis
Diff erent analyses were performed with the data collected using RBio and R software (Bhering, 2017; R Core Team, 2021). A t-test with a 5% signifi cance level for type I error compared the relevance of PBZ application for the eff ect of fi xed treatment. Fixed models were also used to analyze the signifi cance between the treatments. The covariances of matrix structures were assessed to fi t the best model, according to Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC) values.

Flowering analysis
The top graftings performed six months before natural fl owering produced fl ower buds from the fi rst evaluation, which occurred three months after the grafting procedure. However, the count for fl ower buds started seven months after this grafting period ( Table  1). The last assessment, in August 2019, was used to demonstrate the eff ectiveness of the methodology. The number of top grafts with fl owering was counted among those alive per genotype. The ratio of bloomed top grafts and those alive 25 months after grafting was higher than 83%. The genotype with the early fl owering used as a graft is a E. urophylla × E. grandis hybrid (UROGRA3). 25 months after grafting, the genetic material presented a total of 15 top grafts with fl owering (78% of live graftings from this genotype). The fl owering of the top graftings for the genotypes UROGRA1, URO1, and GRA3 was not evaluated due to their precocious graft death. The PBZ was effi cient in promoting fl ower buds, fruit production, and fruit development, with a higher number of top grafts fl ourishing with PBZ than without it over time.
The top graftings performed in October (three months before natural fl owering) were successful in the fi rst fl owering season as two fl ourished in the evaluation three months after grafting. URO1 was the genotype with the earliest fl owering when used as scion. The evaluation in August 2019 (22 months after grafting) found, in most cases, a high ratio of genotypes that fl ourished among those alive. A total of 38 top graftings fl ourished and the highest number was found for the hybrid with PBZ application. The PBZ was effi cient in inducing fl owering in general. The fl owering capacity of the UROGRA1 and GRA3 genotypes was not evaluated, because of the precocious death of the grafts. The other living genotypes fl owered for the fi rst time in diff erent periods.
The number of fl ower buds produced in 2018 was higher for the top graftings carried out in July (Table 1), but the top graftings from the second period stood out. Higher fl ower bud and fruit production were obtained for the materials grafted in October, compared to those grafted six months before natural fl owering. The PBZ induced greater fl ower bud and fruit production. Eight top graftings produced a considerable quantity of fl ower buds and fruits during the last evaluations, surpassing two thousand buds in a single top graft. The maintenance of fruits was satisfactory over time, with 34% for the top grafts done in July and 48% for those grafted in October.
Overall, the fl owering achieved with top graftings of both periods produced enough material to obtain the second cycle of self-pollination and the crosses between diff erent S 1 × S 1 genotypes. The S 2 selfed progenies were also generated at CENIBRA's Hybridization Orchard, using the top grafts from this project. In summary, from all graftings made, 64 top grafts presented fl owering. Thirty-two were selfpollinated and generated fruit and seeds for the next generation. The seeds were germinated and a total of 529 seedlings (possibly S 2 ) were produced, which were sent for microsatellite genotyping to confi rm S 2 selfed generation. A total of 379 S 2 individuals were confi rmed (72% of success). Part of these seedlings were grafted as top graftings to follow the same process and obtain S 3 .

Infl uence of PBZ on survival and canopy area of top grafting
The top grafting's survival and canopy area growth data were tested with and without PBZ application. Diff erences between treatments were assessed at each grafting time using the t-test at (P< 0.05) type I error ( Table 2).
The PBZ increased fl owering in the top graftings. However, the survival and the canopy area increase did not diff er with or without PBZ application in most cases ( Table 2). The growth of canopy area in the top graftings of six months before fl owering was the only trait aff ected by the PBZ application. For this reason, in most casses its application is not necessary to guarantee higher values for the other two traits.

Repeated evaluations analysis
The evaluation of the survival and canopy area aimed to detect diff erences in the development of top grafts in the two grafting periods (six and three months before fl owering). The adequate models were selected based on the lower values of Akaike Information Criteria (AIC) and Bayesian Information Criteria (BIC). Considering survival, the selected model with the best covariance structure was AR and for canopy area, the HCS covariance structure.
Based on the selected models, the Genotype eff ect showed signifi cance (P< 0.05) for the survival of the top graftings performed six and three months before and for the area of top grafts from three months before (Table 3). T-test was not signifi cant for the area of top graftings performed six months before. The Evaluation eff ect was signifi cant for all the hypotheses, so there is a diff erence in the development of top graftings Table 2 -T test for the fi xed eff ect (PBZ), for the two variables analyzed in the evaluations of the top graftings and for the two grafting periods (six and three months before fl owering). The results related to top graftings were separated in the presence and absence of PBZ. The tcal and p-value obtained from the t test were performed using the R software. Tabela 2 -Teste-t para efeito fi xo (PBZ), para as duas variáveis analisadas nas avaliações das enxertias de topo e para as duas épocas de enxertia (seis e três meses antes da fl oração). Os resultados referentes às enxertias de topo foram separados na presença e ausência de PBZ. Os tcal e p-valores obtidos do teste-t foram realizados usando o software R.  Table 3 -P-values for fi xed eff ects of paclobutrazol (PBZ), genotypes and the interaction between evaluations and genotypes. These probability values were generated from the repeated evaluations model under heterogeneous compound symmetry (HCS) covariance matrix structure for Area (m 2 ) and auto regressive (AR) for survival (%), for each grafting period. Tabela 3 -P-valores para efeitos fi xos de paclobutrazol (PBZ), genótipos e interação entre avaliações e genótipos. Valores de probabilidade gerados a partir do modelo de medidas repetidas sob estrutura de matriz de covariância de simetria composta heterogênea (HCS) para Área (m 2 ) e autoregressiva (AR) para sobrevivência (%), para cada período de enxertia.

Traits Survival (%) Area (m²) Treatments
Six months before Three months before Six months before Three months before Genotypes < 0.0001* < 0.0001* 0.4137 < 0.0001* Evaluation < 0.0001* < 0.0001* < 0.0001* < 0.0001* Genotypes × Evaluation < 0.0001* 0.6853 < 0.0001* 0.004* over evaluation time. The Genotype × Evaluation interaction was statistically signifi cant for most scenarios except for the survival in grafting conducted three months before natural fl owering. The overall conclusion demonstrates that considering all the eff ects analyzed, performing grafts of self-pollinated individuals three months before natural fl owering is the most appropriate period for Eucalyptus species, to obtain satisfactory survival and development rates.

Flowering and paclobutrazol infl uence upon top grafting development
The fl owering started three months after grafting with subsequent fruit production, confi rming that self-pollinated genotypes presented early fl owering and retained the successfully produced fl ower buds. The top graftings from bolth periods fl ourished, but the top grafting of six months before natural fl owering (performed in July) can be carried out to obtain short-term results. In this period the top grafting has adequate time to develop and start producing a higher number of fl ower buds in the fi rst year of development after grafting.
The top grafting performed three months before (October) guaranteed greater fl ower bud and fruit production in the second year after grafting. The superior performance of the top graftings performed in this period occurred in the rootstocks with PBZ application. Also, this period's higher long-term survival rate contributed to the higher fl owering production. These top graftings maintained and developed until their second fl owering season, producing more fl ower buds and fruits. Thus, despite requiring a longer period (almost two years after grafting), the three-months before period is recommended, ensuring a higher number of living genotypes and abundant fl owering.
The predisposition and the time needed to start fl owering diff ered in the self-pollinated genotypes used as a scion. This diff erence in plants is complex and involves genetic interactions, environmental factors, and development stages (Ha, 2014;Van Eeuwijk et al., 2019). The physiological aspect, for instance, refers to the plant's ability to transport nutrients, phytohormones, and organic compounds from the root to the crown and vice versa (Nanda and Melnyk, 2018). The affi nity between the scion and the rootstock is also important to ensure the morphological, anatomical, physiological, and biochemical compatibility between them (Tedesco et al., 2022). This aspect was considered in the present methodology since the rootstock is the maternal parent of the genotype that originated the scion material. The success of grafting to induce fl owering also varied for cassava genotypes (Ceballos et al., 2017).
Factors such as environmental adaptation (average temperature and photoperiod) and the expression of proteins aff ects the predisposition of genotypes to the fl owering phenomenon (Turck et al., 2008;Amasino, 2010;Yeoh et al., 2011;Ha, 2014;McClung et al., 2016;Sharif et al., 2021). The Flowering locus T -FT protein is a mobile signal produced in the leaves and transported via the phloem to the apical meristem, where it interacts with other transcription factors to initiate fl oral development (Amasino, 2010;Lee and Lee, 2010;Min and Kramer, 2020). The induction of FT expression in the leaves and its movement towards the apex trigger fl owering (Wigge, 2011;Yeoh et al., 2011). Graft techniques such as top grafting can take advantage of protein mobility to induce fl owering (Wu et al., 2022). This protein is naturally produced and translocated to the top grafts canopy because the rootstock is in an advanced development stage.
The relevance of PBZ is another infl uence on fl owering and the application of this growth regulator induced early fl owering/fruiting in Mangifera indica (Srivastav et al., 2010;Kumar et al., 2021). However, its eff ect varies with species, age, and concentration of phytohormones used . The PBZ application also increases ABA and cytokinin concentrations in Solanum trilobatum. The higher concentration of cytokinin improved the chloroplasts diff erentiation and chlorophyll biosynthesis, preventing the degradation of these pigments (Nivedithadevi et al., 2015).
The fl owering of the top graftings was greater when performed in rootstocks with PBZ due to reduced growth, accelerated maturity, chlorophyll concentration, and greater translocation of photoassimilates caused by this growth regulator (Hajihashemi, 2018). The production and growth of plants depend on photosynthetic rates, essential for plant growth and development. PBZ application guarantees greater photosynthetic yield per unit of Figure 1 -Use of top grafting for inbred lines production. A and B-Graft methodology using twigs collected from self-pollinated individuals; C and D-Paclobutrazol application over the rootstock's root area; E-Top grafting blooming three months after grafting; F and G-Flower bud and fl ower production was satisfactory in many top graftings; H, G, and J-Top grafts were able to provide fruit development until the seed harvest period. leaf area and better partition of the photoassimilates (Xia et al., 2018). Thus, this greater photoassimilate production allowed higher fl ower bud production ( Figure 1).
The choice of branches in rootstock can aff ect the fl owering rates and survival of top graftings. The repetitions were based on four grafts in the same branch. The crown growth of those grafted on the main branches was higher than using the laterals, as the laterals tend to stagnate and die. The eff ect of branch senescence in the survival rate has also been described for Pinus elliottii var. elliottii. Higher graft survival rates were observed in the middle portion of the crown, followed by the apex and lower percentages in the basal portion (Perez et al., 2007). Also, the maintenance of vigor (pruning and fertilization) can contribute to the success of top grafting, guarantying the development of those that live and adequate fl owering production (Almqvist, 2013b). The vigor also favored the grafting survival in young branches of Araucaria angustifolia (Wendling, et al., 2017).
Top grafting also allows the seedlings obtained at each cycle of self-pollination in the fi eld or a nursery (young seedlings selected by the Genome-Wide Selection-GWS) to be grafted, producing the subsequent cycle (Castro et al., 2021). Its use will rule out the exclusive use of traditional breeding programs to obtain superior clones as the inbred lines will be obtained with self-pollination cycles, tending to revolutionize tree improvement. The GWS can also contribute selecting the best individuals of each generation at an early stage for grafting (Lebedev et al., 2020). This will solve the problem of genetic improvement programs, related to the time needed to complete each generation. Thus, the inbred lines can be accomplished by using two diff erent methodologies: top grafting and GWS.
The depression for traits due to inbreeding may aff ect this process as the generations advance, since Eucalyptus species are allogamous (Cobb et al., 2019;Berlan, 2018;Nickolas et al., 2019). However, the eff ect of inbreeding depression on ten commercial Eucalyptus clones showed a small magnitude for circumference at breast height [mean depression of 17.5% (P< 0.05)] and, especially, for wood basic density (Bison et al., 2004). It is expected to avoid the eff ects of inbreeding by selecting and grafting the best self-pollinated individuals per generation, which will make it possible to achieve the high rate of homozygosity expected for the inbred lines produced.

Repeated evaluations analysis
A set of a priori candidate models have been defi ned as possibilities to analyze the survival and canopy area data. The AIC (Akaike, 1974) and BIC (Schwarz, 1978) values were used to compare these models and defi ne the best option based on their results for each scenario. After selecting the HCS model, it was used to analyze the Genotypes and Evaluation eff ects upon the survival and development of canopy area. Only the Genotype eff ect presented no signifi cance for the area trait in top graftings performed six months before natural fl owering. This result indicates no diff erences among development and survival for those top graftings, so it would not need to be measured several times as the result was uniform.
The results of the Genotype eff ect also show that for survival, some genotypes are more likely to remain alive, and either six or three months before fl owering are periods that can be recommended for grafting. In both periods the result was similar for the genotypes that remained alive. However, the temperature fl uctuation is greater in July than October and this factor tends to aff ect the initial bloom and shoot development of Eucalyptus grafts, increasing early mortality. The graft period also infl uences the survival rate for A. augustifolia (Gaspar et al., 2017). Thus, choosing the three months before natural fl owering provides better survival.
The analyses of the Evaluation eff ect demonstrate that recurrent evaluations are necessary to make an eff ective defi nition of the best genotype for grafting made in both periods. The initial evaluations would not make it possible to ranking the genotypes, determining which would present good compatibility and tendency for satisfactory development to continue with the self-pollination cycles. The interaction Genotypes × Evaluations also shows the necessity of monitoring all the development stages of top grafting until fl owering and development of fruits is confi rmed. Therefore, the defi nition of self-pollinated genotypes more conducive to the canopy performance should be done after evaluations confi rming the adequate grafting period for the species.

CONCLUSIONS
The top grafting technique is viable to induce early fl owering of Eucalyptus. The number of fl ower buds and fruits produced was satisfactory in the two grafting periods to obtain the second cycle of selfpollination and the crosses between diff erent S 1 × S 1 genotypes. However, the number of fl ower buds and fruits produced were higher for the period three months before natural fl owering, this being the period recommended for the methodology's execution.
The application of paclobutrazol increased the fl owering of the top grafting but, in general, its application does not aff ect the survival rate and the development of the canopy area of the grafts.
The appropriate choice of genotypes apt for grafting can guarantee the survival of a greater number of top graftings and the highest growth rates in the canopy area. Also, the constant monitoring and maintenance of the graft canopy quality is necessary to ensure its vigor.