Early selection in Eucalyptus camaldulensis Dehnh . progenies in Savanna , Brazil

1 Forestry and Environmental Sciences Graduation Program – UFMT, Cuiabá/MT, Brazil. 2 Department of Forest Engineering, Faculty of Forestry – UFMT, Cuiabá/MT, Brazil. 3 Federal Institute of Education, Science and Technology of Mato Grosso – IFMT, São Vicente/MT, Brazil. 4 Local Development Postgraduate Program – UCDB, Campo Grande/MS, Brazil. 5 Environmental Sciences and Agricultural Sustainability Postgraduate Program – UCDB, Campo Grande/MS, Brazil.


INTRODUCTION
Eucalyptus (Eucalyptus spp.) is the most planted forest tree species in Brazil.Rapid growth, wood production to different uses, adaptability to different soil and climatic conditions, associated with the intense improvement and optimization of silvicultural practices programs, had facilitated the expansion of the culture and its cultivation in the country (Gonçalves et al., 2013).Currently, Eucalyptus species are planted on 5.7 million hectares in Brazil (IBÁ, 2017).Eucalyptus grandis, Eucalyptus urophylla and their hybrids are predominantly cultivated (Pires et al., 2011) but other species like Eucalyptus camaldulensis Dehnh., also has potential for productive plantations.
E. camaldulensis is a species that presents a wood that *Corresponding author.E-mail: paulamartin.bio@gmail.com.
Author(s) agree that this article remain permanently open access under the terms of the Creative Commons Attribution License 4.0 International License is employed in civil construction due to its high density, and also in the manufacture of paper, cellulose, woodbased panels, woodwork and carpentry, transmission poles, firewood, coal and wooden sleepers (Santos et al., 2008;Butcher et al., 2009).This species also has the capacity of growing in critical geographic zones such as in moderately poor soils, and is resistant to long periods of drought and concentrated rain, similar characteristics found in the savanna biome and also in many regions of Mato Grosso State (Azevedo et al., 2015) a state located in the Center-West region of Brazil.Though the Eucalyptus is the forest species most planted in Mato Grosso State (IMEA, 2013), there are no eucalyptus improvement program especially for this State.The reason may be due to the distance from commercial centres and the absence of great forest businesses that work with the genus.Thus, the State's plantations are established with genetic materials not adapted to its soil and climatic conditions.Consequently, it is necessary to seek alternatives to enable the selection of adequate genetic material in a short period.
In early selection model, several selection criteria can be analyzed and this technique is important for breeding programs because of the advantages it presents.Associated with repeatability, this model makes it possible to select with greater certainty the best material to obtain genetic gains in the first years of plant development (Wu, 1998(Wu, , 1999;;Apiolaza, 2009).
In genetic improvement programs of forest species such as the Eucalyptus, the time factor is an important variable, due to its long productive cycle (Moraes et al., 2014b).Hence, the early selection presents itself as the main alternative for reducing the time of the improvement cycle, once it is carried out before rotation age with the aim of minimizing the generation interval, and consequently, increasing genetic gain without losing productivity (Marques Junior, 1995;Fujimoto et al., 2006).
In this context, the present work aimed to assess the possibility of carrying out early selection of potential clones in the first progeny test of E. camaldulensis located in the savanna of Mato Grosso State, Brazil.

Progeny and research area
Progeny test was conducted in the experimental field of the Federal Institute of Education, Science and Technology of Mato Grosso (IFMT, Campus São Vicente) in the municipality of Santo Antônio do Leverger, Mato Grosso State, Brazil.The area is located in the biome savanna, at latitude 15°49'21"S and longitude 55°25'06"W, with 750 m of altitude and 2000 mm of annual average precipitation.According to the Köppen-Geiger classification, the climate of the region is of the Aw type, with average temperature of 20°C.
For establishment of the experiment, seeds of E. camaldulensis were collected from open-pollinated trees at Faculty of Engineering, UNESP Campus Ilha Solteira (provenance Katherine River, Queensland, Australia) in partnership with Embrapa Forests and Federal University of Mato Grosso (UFMT).
The half-sib progenies test was established under randomized block design with 132 treatments (progenies), five repetitions and three plants per plot in single lines at 3 x 2 m spacing.At three and six years old, the circumferences at breast height (in centimeters) and total height (TH; in meters) of all trees of the experiment were measured with the aid of a tape measure and the Blume-Leiss hypsometer, respectively.For data analysis, circumference values were converted into diameter at breast height (DBH), and the values divided by π.

Statistical analysis
In both ages, the estimates of the components of genetic parameters variances were obtained by the restricted maximum likelihood method and best linear unbiased prediction (REML/BLUP) on the basis of unbalanced data and employing the Selegen-REML/BLUP genetic-statistical software developed by Resende (2002).The model used was model 1, which is suitable for half-sibling progeny tests under randomized block design, with several plants per plot, only one site and a single population.The referred model is given by "y = Xb + Za + Wc + e", where y = data vector; b = vector of block effects (fixed); a = vector of additive genetic effects (random); c = vector of plot effects (random); e = vector of random errors; X, Z and W = incidence matrices for b, a and c, respectively.
The effects of genotypes and plots were tested via analysis of deviance through the chi-square test at 1 and 5% of probability, comparing deviance values with and without the respective effect.
In order to aid the assessment of the feasibility of early selection, the genetic correlation between the two measurements was also estimated by Selegen-REML/BLUP.
Estimates of genetic gains with direct and indirect (early) selection were obtained by the employment of the following expressions, cited by Cruz et al. (2004): where, GSi(x) (%) = genetic gain estimated at age "x" by the selection practiced directly on character "i" at the same age; GSi(x,y) (%) = genetic gain estimated at age "x" by the selection practiced directly on character "i" at age "y", with:h 2 a(x) = individual heritability at age "x"; h 2 a(x,y) = individual heritability at age "x" based on the selection of individuals at age "y"; DS(x) = direct selectionselection differential at age "x";DS(x,y) = indirect selectionselection differential at age "x" based on the selection of individuals at age "y".
We obtained the percentage of coincidence between the best individuals in both ages by ranking the 75 best individuals at three years of age, having observed the frequency of these 75 individuals at the age of six.

RESULTS AND DISCUSSION
For the DBH variable, significance was observed for progeny effects at 5% of probability (Table 1), and thus it is possible to infer that there is genetic variability to practice selection on the basis of DBH both at three and six years of age.As for height, there was no significant observation for progeny effects in both ages by the analysis of deviance, which indicates low genetic variation for this variable and lower gain perspective.Because of this, the other analyses (gain estimates and genetic parameters) will not be presented due to their near-zero values.
Several authors have shown the existence of genetic variability in characters associated with growth in genetic tests with eucalyptus species of three years old, both in early selection of clones (Massaro et al., 2010;Beltrame et al., 2012;Moraes et al., 2014b) and early selection in progeny tests (Henriques et al., 2017).The manifestation of this variability in younger ages favors the early selection process allowing the identification of superior genotypes prior to rotation age.
However, besides the fact that there is genetic variability for the selection in early ages to occur, it is important that the best genotypes be kept in the first positions of the ranking, both at rotation age and at the age that one intends to perform early selection.In the present study, the coincidence between the best genotypes in the two ages (three and six years old) was 86%, showing that there was little change between the two ages.Additionally, the genetic correlation for DBH in the two assessed ages was 95.8%, which means that the selection of individuals' superior in DBH at three years old shall be close or similar to the age of six.
The experimental variation coefficient (CV e %) demonstrates that good experimental precision and its estimates in the present work were around 20% (Table 2), values superior to those found by other authors in eucalyptus progeny tests, as in Freitas et al., (2009) at three years old and Moraes et al. (2015) at 36 months old in different localities.However, the genetic variation presented a magnitude close to environmental variation, once the relative variation coefficient (CV r ) that expresses the relation between the experimental variation (CV e %) and the progeny variation coefficient (CV gi %) was close or equal to 1 (3 and 6 years old, respectively) (Table 2), which evidenced a favourable situation for selection (Vencovsky and Barriga, 1992;Cruz et al., 2004).
The observed accuracies (Table 2) were of high magnitude for both ages (values superior to 0.7) (Resende and Duarte, 2007).This result indicates that the selection based on this character is safe by virtue of the high relation between the predicted and real value.
The block narrow-sense individual heritabilities (h 2 a ) estimated in the experiment were superior to those reported by Henriques et al. (2017) in a progeny test of We verified that, in general, heritability values for the two assessed ages were of medium magnitude (Resende and Duarte, 2007); also, they presented low variation with age (Table 2), which indicates that there was no strong change of genotypic variance in relation to the environmental in the two assessed ages.Some authors report that heritability increases with age, which leads to infer that there is a greater influence of environment on juvenile characteristics (Borges et al., 1980;Kalil Filho et al., 1982) or that, insofar as trees become adults, the genotype exerts greater influence on the expression of phenotype (Borges et al., 1980), a fact that was verified by Henriques et al. (2017), who assessed seven years old E. urophylla progenies and Beltrame et al. (2012) with Eucalyptus spp.hybrid clones at three and seven years old.Especially in this study, the maintaining of heritability values in different ages is another factor that favors selection, so that the estimated gains in the two ages have similar values since an early period and the same selection intensity is maintained.
Direct genetic gains with the selection of potential clones were similar in the two ages.This situation was also reported by Pinto et al. (2014) with E. urophylla clones at the ages of three and seven (Table 3).Genetic gain with indirect selection by three years old was 33.65%.Thus, we observe that, when practicing indirect selection and selecting the DBH character by three years of age with the aim to obtain results in the DBH character at six years, the values are similar to those of the direct selection gain, which confirms the possibility of the early selection of progenies at age of three.
Previous experiments with Eucalyptus have shown this relation between selection gains, which provide good estimates for early selection.Massaro et al. (2010), with the prediction of genetic gains by 25 months, 50 months and 72 months in Eucalyptus spp.clones, indicated that genetic gains maintained close values between the ages on the basis of the DBH variable.Araújo (2015) also verified the potentiality of early selection in two clonal tests of E. grandis x E. urophylla hybrids and obtained estimates that prove that the selection of individuals at two years old results in genetic gain in adult age (72 months old).

Conclusion
Existence of genetic variability in the selection for DBH at three and six years of age, associated with the closeness of the estimated indirect gain values with selection at three years old, and the high correlation between genotypic values for DBH in the two assessed ages, indicate that the process of selection at an age inferior to that of rotation proved to be viable, not compromising gains and identifying the best genetic materials early.These results will subsidize the first breeding program of E. camaldulensis in the Mato Grosso State, Brazil.

Table 1 .
Analysis of deviance (ANADEV) for diameter at breast height (DBH) and total height (TH) in progenies of Eucalyptus camaldulensis at three and six years of age.

Table 2 .
Genetic parameters for DBH (diameter at breast height in centimeters) at 3 and 6 years of age from progeny test of E. camaldulensis.
a: Block narrow-sense individual heritability, that is, additive effects; h 2 mp: Heritability of the progeny mean; Acprog: Accuracy of progeny selection; CVgi%: Coefficient of variation of progenies; CVe%: Coefficient of experimental variation; CVr%: Coefficient of relative variation.

Table 3 .
Direct and indirect gains and selection mean for the 75 best individuals (potential clones) in progeny test of E. camaldulensis.found by Rocha et al. (2006) with progenies at 58 months of age and Moraes et al. (2014a) who assessed at the ages of 24, 36, 48 and 60 months in clonal tests of Eucalyptus spp., where the values were superior to 0.7.