Coffee production has contributed significantly to economic and social development in Brazil and is of great importance to Brazilian agribusiness. Brazil has been the world's largest producer and exporter of coffee for over 150 years (Paiva et al., 2010). National productivity in 2018 was 1903.2 kg per hectare. In 2019, 1509.6 kg per hectare was estimated (CONAB, 2019).

The Brazilian coffee industry has undergone significant changes as crop has moved into the Cerrado areas, particularly in its production system (Oliveira et al., 2010). The Cerrado produces excellent quality coffee due  to  its two well-defined seasons: rainy summer and dry winter (Fernandes et al., 2012). In addition, controlled water stress can be used to standardize the flowering and ripening of fruits in the Cerrado (Guerra et al., 2005).

Brazil has 131 registered cultivars of Coffea arabica L. However, not all are able to adapt to different growing conditions and reach their productive potential. Botelho et al. (2010) point out that genotype with superior behavior in a certain environment may not behave satisfactorily under other conditions. Thus, it is necessary to improve and select genotypes to ensure that they express desired traits. The development of new cultivars is achieved through genetic improvement processes (Paiva et al., 2010), which require genetic variability in the population (Ivoglo et al., 2008). Evaluating progenies in several locations is an important step in the final phase of a plant breeding program. With this information, the interaction between genotype and environment can be determined. Additionally, when interactions exist, subsidies can be provided to encourage cultivation at specific sites (Pinto et al., 2012). By characterizing genetic divergence, efficiency in the selection of parents in breeding programs can be increased (Silva et al., 2013). In this process, hundreds to thousands of individuals are evaluated to identify superior and divergent genotypes for certain characteristics in order to design by recombination (Silva et al., 2016). The dissimilarity analysis is used to quantify genetic variability and the relative contribution of the variables to the genetic dissimilarity, allowing for the identification of promising combinations (Torres et al., 2015).

The genetic dissimilarity allows one to identify promising genotypes for breeding programs and to recommend for cultivation. Giles et al. (2019) verified genetic divergence among 34 genotypes of Coffea sp. and conclude that phenotypic variations occurred predominantly due to genetic causes. Thus, the objective was to study the genetic dissimilarity among 35 genotypes of C. arabica in the Cerrado, under drip irrigation, using phenological data and multivariate statistics.

 

The experiment was conducted at the Ceres Campus of the Federal Institute Goiano, GO. The Ceres Campus is located in the center of Goiás in the São Patrício Valley (UTM: E = 649,582.00 m and N= 8,302,194.00 m), and is characterized by having flat relief, very deep eutroferric red nitosol, clay texture, and an altitude of 556 m. The climate, according to the classification of Köppen, is Aw type (tropical climate with wet and dry seasons- Tropical Seasonal, dry winter), with an average annual temperature of 25.4ºC (average minimum: 19.3°C; average maximum: 31.5°C). The annual precipitation is approximately 1700 mm.

The experiment was conducted on April 8, 2015 in a randomized complete block design. The experiment consisted of 35 treatments, 31 cultivars and 4 progenies (Table 1) with four replication and 10 plants that were placed 3.50 × 0.75 m apart. The eight central plants were considered for analysis. During the experiment, recommended management practices as fertilization, phytosanitary management and irrigation for the crop were followed. In the dry season, drip irrigation occurred on Mondays, Wednesdays, and Fridays to account for the need for the crop (Kc) and evapotranspiration in a class A tank. Fertilization was performed based on soil analysis results and recommendations of the 5^th approximation of the Soil Fertility Commission of the State of Minas Gerais (Guimarães et al., 1999). At 6, 12, 18 and 24 months after planting, stem diameter (DST), canopy diameter (DCA), plant height (HEI), number of orthotropic branch nodes (NOBN), average length (cm) of orthotropic branch internodes (ALOBI), total number of nodes at the 2nd plagiotropic branch pair (TNPB), total length (cm) at the 2nd plagiotropic branch pair (TLPB), and average length (cm) of plagiotropic branch internodes (ALPBI) were measured.

 

 

Data were analyzed using Analysis of Variance, F, and the Scott-Knott test at 0.05 of means for phenological parameters. Hierarchical Agglomerative Cluster (HAC) was used to examine dissimilarity by measuring average Euclidian distance. Additionally, Ward’s agglomeration method was used to obtain dendrograms and the Pearson’s method (n) for Principal Component Analysis (PCA) was used to obtain the correlation matrix and distance Biplot. Statistical analyses of genetic data were performed using the software XLSTAT 2014.5.03. The number of groups in the dendrogram was determined by the automatic truncation function, which attempts to create homogeneous groups (XLSTAT-MX, 2005).

 

 

Coffee genotypes showed differences in phonological variables at 6, 12, 18, and 24 months after planting (Tables 1 to 4). Evaluations that occurred at 6 and 18 months after planting coincided with the end of the dry season, while evaluations that occurred 12 and 24 months after planting coincided with the end of the wet season. Temporal variability was observed in genotype behavior, as the growth of each material to diverse edaphoclimatic conditions differed among evaluations.

 

 

 

 

Meireles et al. (2009) state that various phenological phases of C. arabica are affected by environmental conditions, especially by photoperiodic variation and meteorological conditions (rainfall distribution and air temperature). In this experiment, the evaluations at 6 and 18 months after planting, in month October, of season rainy beginning and the photoperiod increasing, peaking in December. At 12 and 24 months after planting, month of April, the end of the rainy season and the photoperiod with short days, with minimum in June. Genetic diversity was observed between genotypes in the adaptability and interaction of the genotypes with the environment, so multivariate techniques were used to evaluate genetic divergence.

The 35 genotypes were clustered into three groups at each evaluation using Hierarchical agglomerative cluster analysis (Figures 1 to 4). Differences in genotypes were observed among groups for each evaluation. At 6 months, the most distant group consisted of Yellow Catucaí 2SL, Araponga MG 1, Sacramento MG 1, 23 II, Yellow Catucaí 20/15 pit 479, Sarchimor MG 8840, IBC-Palma 2, and New Acauã. At 12 months, the most distant group consisted of Yellow Catucaí 2SL, Asa Branca, Sacramento MG 1, and Sarchimor MG 8840. At 18 months, the most distant group consisted of Yellow Catucaí 2SL, Tupi IAC 1669-33, 23 II, Red Obatã IAC 1969-20, Sacramento MG 1, and Sarchimor MG 8840. At 24 months, Yellow Catucai 2SL was distinct from the other genotypes.

 

 

 

 

 

The first group formed in the 24^th month dendrogram consisted of the treatments 34, 4, 14, 2, 7, 6, 26, 8, 27, 1, and 30 (Figure 4). This group had the lowest averages, particularly for DST, DCA, HEI, and TLPB. The second group consisted entirely of the genotype Yellow Catucaí 2SL. The third group consisted of the treatments 16, 35, 10, 12, 32, 9, 11, 19, 5, 24, 13, 17, 31, 3, 25, 29, 18, 20, 33, 15, 28, 21, and 22 and had the lowest averages for HEI and ALPBI, with only TNPB having a greater amount of significant averages (Table 4).

PCA results showed that the relative contribution of phenological variables to genetic dissimilarity (frequency) was 78.85% at 6 months (Figure 5), with the F1 component contributing 53.80% and the F2 component contributing 25.05%. For the second evaluation period (12 months), a frequency of 81.38% was observed (Figure 6) with F1 contributing 57.99% and F2 contributing 23.39%. At 18 and 24 months, the relative contribution was 79.73% (F1 = 47.48% and F2 = 32.25%) and 81.97% (F1 = 47.48% and F2 = 32.25%), respectively (Figures 7 and 8). Thus, variability in the contribution   of    phenological    variables    was     observed mainly at 18 and 24 months. This could be due to a decrease in photo-assimilated reserves, causing a decrease in the growth rate of the plants, as they were in the process of filling the grains, which is considered to be a substantial photo-assimilates drain (Arantes et al., 2006).

 

 

 

 

 

 

 

 

 

 

 

The authors have not declared any conflict of interests.