Genetic variability of conilon coffee population from cultivar ‘ES8152’ based on morphoagronomic variables

This study aimed to analyze the genetic variability of Coffea canephora population with 190 genotypes from cultivar ‘ES8152’, based on morphoagronomic characteristics and vegetation index, to identify the most important characteristics for genetic divergence and compare them with commercial clones. The experiment was installed, in 2019, at the Bananal do Norte Experimental Farm/INCAPER, Cachoeiro de Itapemirim, ES, Brazil. The experiment was carried out in Federer’s augmented block design with three blocks, four common treatments (commercial clones A1, LB1, V8 and V12) and 190 regular treatments, genotypes from the seed production field of the conilon coffee cultivar ‘ES8152’. At 24 months of age 14 morphoagronomic characteristics and vegetation index were evaluated. Descriptive analysis of the data, the estimation of the Standardized Euclidean Distance (ED) followed by the group ing by the methods of Tocher, UPGMA and principal coordinates, in addition to the relative importance of the characters estimated by the Singh meth odology were performed. The most distant genotypes were 62 and 83 (ED=2.620) and the closest were 42 and 160 (ED=0.208). Genotype 83 stood out as the most distant among the others. The optimization and hierarchical groupings allowed the identification of genotypes 15, 81, 107 and 184 as similar to commercial clones. The discard analysis of variables recommended the elimination of the vegetation index and average internode length of the next diversity analysis. Principal coordinate analysis found phenotypic similarity of the genotypes 30, 81, 115, 141 and 163 with the clone V12, of the genotype 119 with the clone A1 and genotype 17 with clone LB1. The study, of morphoagronomic characters, allowed to detection the genetic diversity existing in the materials evaluated, indicating those with phenotypic similarity with the commercial clones, being possible the early identification of promising genotypes, agronomically superior, to start a breeding program for clonal selection, recurrent selection and controlled crosses to maximize heterosis.


INTRODUCTION
Coffee is one of the great products of brazilian agribusiness with expected production of 48,807 thousand bags (60 kg) in an area of 1,824.7 thousand hectares in 2021 (Companhia Nacional de Abastecimento -CONAB, 2021). The Espírito Santo state stands out on the national scenario as the second largest coffee producer, with great emphasis on the conilon coffee. The estimated cultivation area, in Espírito Santo state for 2021, was 400.4 thousand hectares with a production of 14.126 million bags, with 2.986 million bags of arabica coffee and 11.140 million bags of conilon coffee (CONAB, 2021). Major technological advances in Capixaba coffee growing are related to research and rural extension work, carried out by different multidisciplinary and multiinstitutional teams. In this context, the clonal (nine) and seed propagated (two) cultivars obtained and made available by the Capixaba Institute of Research, Technical Assistance and Rural Extension (INCAPER) were registered, which constitute the basic genetic material of the planted and renewed crops.
Currently, the main cultivar of C. canephora of seminal propagation is 'ES 8152', also known as 'Conquista', developed by INCAPER and released in 2019. It is a cultivar formed by the recombination of 56 agronomically superior genotypes, in an isolated, for seed production. The main characteristics of this cultivar are broad genetic base, rusticity, adaptation, high productivity, production stability, drought tolerance, rust resistance, high vegetative vigor, good grain processing yield. In addition, it is recommended mainly for small and medium family-based producers (Ferrão RG et al., 2019).
The planting of cultivars propagated by seeds ensures greater natural variability of the species in crops, being considered a simple method and the main strategy for obtaining highly heterozygous offspring (Souza et al., 2021). Conilon coffee (Coffea canephora) is an allogamous and diploid species, with gametophytic self-incompatibility (Moraes et al., 2018;Tran et al., 2017). Thus, reproduction by seeds results in a highly diverse population, in which each plant may differ from the others in terms of architecture, shape and size of grains and leaves, maturation pattern and susceptibility to biotic and abiotic stresses (Dubberstein et al., 2020). The genetic diversity existing among and within populations can be measured by the difference among the phenotypic values of their accessions and is obtained in field experiments using a considerable number of morphological, agronomic and other characteristics (Pontes et al., 2020).
The evaluation of genetic diversity allows the identification of distinct heterotic groups, which, when oriented in controlled crosses, will allow the obtainment of offspring with higher heterosis. Several multivariate methods can be applied for this purpose, such as dissimilarity measures calculated from the Euclidean distance or generalized Mahalanobis distance; clustering methods such as hierarchical (UPGMA) and optimization (Tocher); and dispersion techniques (Cruz;Regazzi;Carneiro, 2012).
Face to importance of cultivars propagated by seeds to the genetic improvement program, this study aimed to evaluate the genetic variability of 190 genotypes from conilon coffee cultivar 'ES8152', propagated by seeds, based on morphoagronomic characteristics and vegetation index, comparing them with commercial superior clones, in addition to identifying the most important characteristics for genetic divergence.

Genetic materials
The genotypes came from the seed production field of the cultivar 'ES8152', composed of 56 superior clones, from the INCAPER improvement program, planted in row and in an isolated field. The seed field was implemented in 2016/17 at the Bananal do Norte Experimental Farm (FEBN), belonging to the South Center for Research, Development and Innovation (CPDI Sul)/INCAPER in Pacotuba, district of the municipality of Cachoeiro de Itapemirim, Espírito Santo, Brazil. In the first harvest (2019), a seed sample was collected from all plants. These samples were mixed to form a composite sample. Part of the sample (population) was taken to a nursery for the production of seedlings, together with the four clonal witnesses.

Experimental description
The experiment was installed in November 2019 at FEBN, which is located in the south of the Espírito Santo state, Brazil, at latitude 20º45' S, longitude 41º17' W and altitude of 140 meters. The soil is classified as dystrophic Red-Yellow Latosol, Cwa climate with rainy summer and dry winter according to Köpen's classification, mean annual precipitation of 1,200 mm, mean annual temperature of 23 °C and undulating topography.
Experimental design adopted in this study was Federer's augmented block (Federer, 1956) design with three blocks, 240 regular treatments (genotypes) and four common treatments (clonal witnesses), at the spacing of three meters between lines and one meter between plants. The fertilization of planting and conduction followed the fertilization and liming manual for Espírito Santo state (Prezotti et al., 2007) and the cultural and phytosanitary treatments were carried out accordingly with the requirement of the crop, following the current recommendations for conilon coffee (Ferrão RG et al., 2019).

Agronomic evaluations and vegetation index
At 24 months of age, 190 genotypes of the 240 of the experiment were evaluated, in addition to the four clonal witnesses, with respect to the following characteristics: CH: Canopy height was measured with a graduated ruler, by the distance between the beginning of the coffee tree canopy and its end (cm); PH: Plant height was measured with a graduated ruler, by the length of the largest orthotropic branch (cm); LLB, LMB and LUB: Length of the lower, middle and upper plagiotropic branches, respectively, were measured with a graduated ruler, by the largest plagiotropic branches in the lower, middle and upper portions of the coffee tree (cm); IL: Average internode length of the largest orthotropic branch, was estimated by the ratio CH/NN (cm); SBD: Stem base diameter was measured with a precision digital caliper (0.01 mm) in the intermediate position of the soil up to the first plant node perpendicular to the planting line (mm); CD: Canopy diameter of the coffee tree was measured with a graduated ruler, by the length of the greatest distance of the end of the branches that compose the coffee tree canopy, perpendicular to the planting line (cm); NDVI: Normalized Difference Vegetation Index was measured with a PlantPen NDVI-300 portable sensor (Photon Systems Instruments PSI, Drásov, Czech Republic), using two leaves of the third or fourth pair, from the plagotropic branch, from the middle position of the plant (und); NLLB, NLMB and NLUB: Number of leaves from lower, middle and upper plagiotropic branches, respectively, were obtained by counting the number of leaves in the branches in which LLB, LMB and LUB were evaluated (und); NF: Number of fruits in the plant was obtained by counting the fruits present in the plant from plagiotropic branches used to evaluate LLB, LMB and LUB (und). NN: Number of nodes was obtained by directly counting nodes in the largest orthotropic branch (und); NPB: Number of plagiotropic branches, was obtained by directly counting the number of plagiotropic branches of the coffee tree (und).

Statistical analysis of data
Descriptive analysis of the original data was performed to estimate the phenotypic pattern of cultivar 'ES8152' in comparison with commercial clones.
Based on the set of characteristics (15) and genotypes (194), the matrix of genetic distances was estimated using the Standardized Euclidean Distance (ED), and the grouping of genotypes by the Tocher optimization methods (Rao, 1952), hierarchical UPGMA and principal coordinate analysis. The cophenetic correlation coefficient between the graphical distance matrix and the original distance matrix was estimated. And the relative importance of characters in genetic divergence was estimated by the methodology proposed by Singh (1981). In the principal coordinates analysis, the number of axes necessary to preserve at least 60% of the variance of the original dataset was used.
Statistical analyzes were performed in the computer applications R (R Core Team, 2019) and GENES (Cruz, 2016).

RESULTS
Descriptive analysis of the data (Table 1) demonstrates that the mean and median of the evaluated characteristics are close and, therefore, the data are close to the normal distribution, except for the NF characteristic, in which the median was zero. Several genotypes did not show production  at 24 months of age due to the longer period of juvenility of seminal seedlings compared to clonal seedlings. A higher standard deviation and data variation index were observed for the NF characteristic, demonstrating a greater variability in the observations. The means of commercial clones showed values between the maximum and the minimum observed in the data set under study and clones A1, LB1 and V8 presented the highest values for NF. The observed mean of clone V12 was lower than the mean of the data set for the characteristics LLB, LMB, LUB, IL, CD and NLUB. Clone V8 mean was lower than the data set mean for IL and NLUB characteristics. The mean of clone LB1 was also lower for the IL characteristic and equal to the mean of the NDVI standard of the dataset.
Based on the genetic distance matrix, the largest and smallest statistical distances were, respectively, between the pairs of genotypes 62 and 83 (2.620) and 42 and 160 (0.208) ( Table 2). In the grouping of genotypes by the Tocher Optimization method (Table 3) and UPGMA (Figure 1), genotype 83 also stood out as the most divergent.
The relative importance analysis of the characteristics for genetic divergence (Table 4) shows that CD, NF, NLLB, NPB, CH and PH accumulated the largest proportion of variability with 21. 979, 18.044, 13.856, 11.061, 10.086 and 9.889 of the variation, respectively. These characteristics added up to about 75% of variation in the data set, thus being the most important in the selection process. The NDVI and IL characteristics presented zero and 0.021 of data variability, being recommended to discard these analyzes in future diversity studies.
The principal coordinate analysis (Table 5) shows the distribution of genotypes in a two-dimensional plane. Due to the accumulation of 64.7% variation in the first three axes, the demonstration of the dispersion of individuals combining the three axes was adopted (Figure 2). It was found that in the three possible combinations, the most distant accesses were distributed at the ends of the graph and the closest ones in the interior. In the axes combination one and two, genotypes 20, 25, 50 and 119 were located close to clone A1, genotypes 17 and 80 were found around LB1, genotypes 30, 34, 81, 115, 124 and 141 close to V12 and genotypes 28, 76 and 135 close to V8. In the axes combination one and three, genotypes 70, 119 and 135 were found close to clone A1. Clones LB1 and V8 were close and neighbors to genotypes 17, 50 and 178, clone V12 was adjacent to genotypes 30, 32, 49, 117, 144, 163, 169 and 171. In the axes combination two and three, clone A1 was close to genotypes 46, 59, 101, 119 and 183, clone LB1 next to genotypes 17, 27, 88, 89, 120, 152 and 182, clone V12 next to genotypes 30, 92, 104, 112, 115, 141 and 163 and clone V8 next to genotypes 19, 74, 86, 110, 114, 145 and 150.

DISCUSSION
The Variation Index values (Table 1) were close to those estimated by Senra et al. (2020) in the evaluation of 38 traits from 323 accessions of the active germplasm bank of the INCAPER at FEBN, demonstrating similar experimental control, with good precision in the evaluation of morphoagronomic characters, at 24 months of age, of genotypes of cultivar 'ES8152' and clonal witnesses. The SBD means obtained in this study were close to those of the active germplasm bank of the INCAPER (Senra et al., 2020), and higher for the characteristics CD, LLB, LMB, LUB, NLMB and NLLB.
The greatest statistical distance observed was greater than those estimated in previous studies based on phenotypic and, or, molecular data from the Incaper research program (Ferrão et al., 2009;Ferrão et al., 2017;Ferrão LFV et al., 2019;Ferrão et al., 2021;Fonseca et al., 2006;Giles et al., 2018;Senra et al., 2020;Souza et al., 2021), demonstrating the presence of genetic variability to be explored. The genetic materials identified as divergent, in the UPGMA, Tocher and principal coordinates clusters, must be monitored for their agronomic characteristics over at least four harvests for selection and use per se and, or, in hybridizations in the INCAPER breeding program. The obtaining these results, in the initial phase of experimentation, are important to guide selection and recombination strategies focused on the formation of heterotic groups for controlled crosses. Additionally, the direct selection of superior genotypes is a fundamental strategy in the composition of trials to determine the value of cultivation and use (VCU), followed by the launch of new cultivars. According to Borém and Miranda (2013) and Ferrão MAG et al. (2019), crosses involving genetically different parents can produce a high heterotic effect and the breeder can obtain hybrids or clones superior to the paternal.  The cophenetic correlation coefficient was high (0.719) and higher than that estimated by Ferrão et al. (2021) with a study of 562 genotypes of the active germplasm bank, demonstrating low distortion between the matrix of graphic distances and real distances.
Based on the results of the relative importance of the characters for genetic diversity, it is recommended to discard the NDVI and IL characters, based on the evaluation of the first harvest. However, recent studies have used the vegetation index in research on coffee cultivation as the relationship between meteorological variables and vegetation index (Mota et al., 2020), the different phenological phases of coffee (Nogueira;Moreira;Volpato, 2018), monitoring of coffee maturation to determine the best moment of the harvest (Nogueira et al., 2021). In addition, Silva et al. (2021) suggest the use of NDVI, as an auxiliary character, in the selection of coffee genotypes from six months of age.
The principal coordinates analysis corroborated with the confirmation of the proximity of genotype 81 with clone V12, because the analysis of axes one and two combined with the UPGMA grouping. In the analysis of the three axes, the stability of proximity of genotypes 30, 115, 141 and 163 and clone V12, genotype 119 with clone A1 and genotype 17 with LB1 was observed.
Studies of this nature are important for the genetic improvement of coffee, as they allow the early identification of promising genotypes, agronomically superior, to initiate a program of clonal selection, recurrent selection and guide controlled crosses to maximize heterosis.

CONCLUSIONS
The study, of morphoagronomic characters, allowed to detection the genetic diversity existing in the population of 190 genotypes from conilon coffee cultivar 'ES8152', propagated by seeds, indicating those with phenotypic similarity with the commercial clones.
The genotypes 62 and 83 stood out with the largest genetic distance, genotypes 42 and 160 as the most similar and the genotype 83 as the most divergent in the study.
The characters canopy diameter, number of fruits and number of leaves from lower plagiotropic branches were the ones that most contributed to genetic diversity.