Multi-trait selection and the relationship between sensory analysis and post-harvest variables in <i>Coffea arabica</i>

Moreira, Daniele Birck; Silva, Sabrina Alves da; Costa, Weverton Gomes da; Caixeta, Eveline Teixeira; Nascimento, Moysés; God, Pedro Ivo Vieira Good

doi:10.1590/1678-992X-2025-0020

ABSTRACT

We aimed to estimate genetic parameters and multivariate selection in coffee related to quality and postharvest practices. In a population of 59 progeny of Coffea arabica, we analyzed the most important traits related to quality. Our analysis reveals significant genetic variance in all traits. We examined the effectiveness of direct selection and multivariate genetic gain using factor analysis. Our results suggest that genetic gains are achievable by all traits, although to varying degrees. In particular, selection based on the sensory quality factor alone can lead to gains opposite to those desired for traits such as ripening uniformity, bean size, and yield. This result may be due mainly to the way postharvest variables are processed for the standard Specialty Coffee Association of America (SCAA) beverage quality tests. Therefore, under this scenario, selection using factor analysis must be applied with care. These results provide a basis for future genetic selection strategies that will improve the quality of coffee.

Keywords:
genetic improvement; beverage quality; factor analysis; correlation network

Introduction

Coffee (Coffea arabica L.), one of the world's most widely consumed beverages, owes its popularity to its unique sensory properties. The quality of coffee is influenced by a complex interplay of genetic and environmental factors, as well as post-harvest practices. The growing specialty coffee market has driven the global coffee sector to enhance its agricultural practices to improve sensory quality.

There is a strong focus on selecting coffee genotypes based on beverage quality, which is influenced by numerous variables (Montagnon et al., 1998). Productivity, bean size, and weight have a direct impact on profitability and are highly valued by producers. Additionally, the uniformity of fruit maturation affects both the sensory quality and the efficiency of post-harvest processes. Therefore, traits related to productivity, ripening uniformity, bean size, and the ratio of processed coffee should be considered when selecting superior genotypes (Montagnon et al., 1998; Dessalegn et al., 2008; Kathurima et al., 2009; Reyes González et al., 2016; Sobreira et al., 2016; González et al., 2019).

Understanding the genetic basis of these traits can aid in breeding superior varieties. Evaluating the correlation between different traits and testing selection strategies is crucial to improving quality and accelerating the development of new cultivars (Ferreira et al., 2005; Paixão et al., 2022). The use of factor analysis in multivariate selection can optimize this process by summarizing many traits into a few latent variables or factors, thereby preserving essential information (Granate et al., 2001; Cruz et al., 2012; Barbosa et al., 2019). Factor analysis enables the structuring and simplification of a set of original data, allowing for the representation of a relatively large number of traits as latent variables or factors, thereby preserving as much information as possible (Cruz et al., 2012). As a result, multivariate selection can be more efficient when applied to a few factors that represent several strongly correlated original traits (Granate et al., 2001; Barbosa et al., 2019). Obtaining interpretable factors demonstrates that correlated variables of interest can be summarized through common factors (Ferreira et al., 2005).

In this context, this study applies factor analysis to identify key factors influencing post-harvest and beverage quality traits in C. arabica genotypes. By analyzing multiple traits simultaneously, we aim to improve the efficiency of genotype selection and predict genetic gains, thereby contributing to the development of superior coffee varieties with enhanced sensory characteristics and post-harvest qualities.

Materials and Methods

Genetic material and phenotypic data

Fifty-nine progenies of C. arabica cultivated at Rio Paranaíba, Minas Gerais state (19°12’59" S, 46°13’57" W, altitude 1124 m) were evaluated in the 2020 harvest season (Table 1). The experiment was conducted in a randomized complete block design with three replications. Each plot comprised ten plants, and the four central plants were evaluated in bulk. A total of 22 variables were evaluated during the harvest and post-harvest process.

Thumbnail

Table 1
Identification, description and genealogy of Coffea arabica genotypes at Rio Paranaíba, Minas Gerais state, 2020 harvest season.

Before harvesting, the °Brix (BRI) was determined by direct reading on a refractometer at 20 °C, using four randomly selected cherry fruits from each plot. All the fruits from the selected plants were harvested, and the yield (YLD) was estimated in liters per plot (L plot^–1). After that, a 20-liter sample was taken for the post-harvesting processes. Variables related to fruit ripening were assessed. These included the percentage of ripe fruit (RIF), unripe fruit (UNR), semi-ripe fruit (SEM), and overripe fruit (OVF) in a representative volume of 1 L per plot. The percentage of floating fruits (FLO) in the 20 L sample was determined by the water immersion method in which the buoyant fruits with lower density were collected and their volume measured. Next, a subsample of 5 L of cherry fruits was separated from the total 20 L sample and dried on raised beds until reaching a moisture content of 11.5 %. We used this specific subsample for subsequent processing.

After drying, the coffee was hulled, and the following measurements were taken: estimated volume of cherry coffee per 60 kg bag (L bag^–1) (LBG), weight of 100 beans (WEI), percentage of defect (DEF) beans, and variables related to bean size: percentage of beans retained on the bottom sieve (B15), sieve 15 (S15), sieve 17 (S17), and sieve 17 together with sieve 19 (T19).

A 300 g sample of coffee from sieves 15 and above was prepared for sensory evaluation based on the protocol of the Specialty Coffee Association (SCAA, 2015). Three Q-Graders, in a blind analysis, assessed the sensory attributes of the beverage: final score (FIN), fragrance/aroma (ARO), flavor (FLA), acidity (ACI), body (BOD), overall (OVE), balance (BAL), and aftertaste (AFT). The scale for the final score ranged from zero to 100, while for the other attributes, it ranged from six to ten. The scores for sweetness, uniformity, and clean cup received the same maximum score and were excluded from the data analysis.

Statistical analysis

Phenotypic data were subjected to univariate analysis of variance (ANOVA) to test the significance of genotypic variance and estimation of genetic parameters. Phenotypic ( $({\hat{σ}}_{p}^{2})$ ), genetic ( $({\hat{σ}}_{g}^{2})$ ), and environmental variance ( $({\hat{σ}}_{e}^{2})$ ) components were estimated via the Method of Moments, based on the mean performance of progenies.

Other genetic parameters were obtained from:

$h^{2} = \frac{{\hat{σ}}_{g}^{2}}{{\hat{σ}}_{p}^{2}} (heritability);$ ;
$\hat{r_{g}} = \sqrt{{\hat{h}}^{2}} = (selective accuracy);$ ;
$C V_{g} (%) = \frac{\sqrt{{\hat{σ}}_{g}^{2}}}{\bar{X}} \times 100^{(coefficient of genotypic variation);}$ ;
$C V_{e} (%) = \frac{\sqrt{{\hat{σ}}_{e}^{2}}}{\bar{X}} \times 100^{(coefficient of environmental variation);}$ ;

and $C V_{g} C V_{e}^{- 1}$ ratio.

After estimating the phenotypic means, we applied Factor Analysis (FA) to investigate the relationship between variables and identify latent variables associated with quality coffee. The FA model was used according to Eq. (1):

(1)

X_{j} = I_{j 1} F_{1} + I_{j 2} F_{2} + \dots + I_{j m} F_{m} + ε_{j}

where X_j is the j-th variable, with j = 1, 2, … v, I_jk, the factor loading for the j-th variable, associated with the k-th factor, and where k = 1, 2, … m; F_k is the k-th common factor and ε_j the specific factor associated to the j-th variable.

The number of factors retained in the model was defined by analyzing the highest eigenvalues and the percentage of accumulated variance. The final factor loadings were established through varimax rotation and used to interpret the factor meanings related to coffee quality complexes. In connection with this, we examined the relationship between phenotypic mean variables using the Pearson coefficient, visualized through network correlation plots (Rosado et al., 2017; Rosado et al., 2019). We used a cut-off value where only lines with |r_ij| ≥ 0.5 were highlighted proportional to the correlation intensity (Epskamp et al., 2012).

After identifying new latent variables (factors), we estimated factorial scores for all genotypes to simulate multivariate selection. Previously, we had calculated the direct gain for each variable using 15 % selection intensity. We decided to increase the factor scores for the variables RIF, YLD, LBG, BRI, S17, T19, FIN, ARO, FLA, ACI, BOD, OVE, BAL, and AFT, and to decrease them for the characteristics UNR, SEM, OVF, FLO, DEF, B15 and S15. In factorial multivariate selection, we used the factor score values as a new variable for a selection criterion to obtain simultaneous gains in the original variables. We obtained genetic gain estimates according to the Eq. (2):

(2)

Δ G (%) = \frac{[100 (D S \times h^{2})]}{\bar{X}}

where ΔG (%) is the percentage selection gain, DS the differential selection, h² the heritability, and $\bar{X}$ the original mean of the trait.

Statistical analyses were carried out using the Genes program (Cruz, 2013) and the "qgraph" package in the R software program (R Core Team, v. 4.1.3).

Results

Significant genetic variance was found in all investigated traits (p < 0.01), except for ARO (Table 2). The estimated heritabilities varied widely. For the sensory attributes, heritability values ranged from 19.47 % for the ARO to 71.25 % for the FIN. It should be noted that the lower heritability for fragrance/aroma is due to the reduced genetic variance.

Thumbnail

Table 2
Genetic and environmental parameters of the traits evaluated in Coffea arabica progenies: F-value, p-value, heritability, accuracy, phenotypic variance (

{\hat{σ}}_{p}^{2}

), environmental variance (

{\hat{σ}}_{e}^{2}

), genetic variance (

{\hat{σ}}_{g}^{2}

), coefficient of genetic variation (CVg %) and the ratio of the coefficient of genetic variation to the coefficient of environmental variation (

C V_{g} C V_{e}^{- 1}

).

The other quality traits of coffee beans, such as bean size and fruit ripening uniformity, exhibit high heritability (Table 2). The sieve variables S15, S17, and P19 had estimates above 70 %. As regards the characteristics of the beans, heritability ranged from 45 % (°Brix) to 64 % (WEI). The yield showed a value equivalent to 55 %. For variables related to fruit ripening uniformity, values ranged from 39 % (overripe fruit) to 65 % (green bean).

The estimated accuracies for all traits showed moderate to high values except for the fragrance/aroma variable. For the attributes assessed in the sensory analysis, accuracy values were also high, exceeding 70 %. The final score variable had the highest accuracy value (84 %), indicating good precision in estimating genetic merit. As regards the $C V_{g} C V_{e}^{- 1}$ ratio, values less than one were found for almost all traits studied, indicating a high influence of environmental factors.

Factor analysis revealed the formation of four groups of latent variables (Table 3). For the composition of the factors, we adopted as a criterion a factor loading above 0.5 and a communality above 0.6. Thus, certain variables in Table 3, although listed, do not exert a decisive influence on the interpretation of the factors, such as the variables B15 and WEI (in the sieves factor) and YLD, SEM, and DEF (in the fruit ripening uniformity factor).

Thumbnail

Table 3
Final factor loadings of the first four factors and communality (C) obtained by factor analysis and varimax rotation. Variables in each factor were selected based on factor loadings > 0.5 and communality > 0.6. The highlighted numbers in bold meet the criteria for each factor. The name of each factor was chosen based on its agronomic significance.

The first factor is associated with the sensory attributes of beverage quality. It has been named the sensory quality complex, encompassing attributes such as FIN, FLA, ACI, BOD, OVE, BAL, and AFT. The second factor is related to bean size variables and is referred to as sieves (S17, S15, T19). The third factor is influenced by variables related to ripening and harvest practices, which explains the complex called fruit ripening uniformity (RIF, UNR, BRI). The fourth group, the overripe fruit factor, is similar to the third group and is represented by traits OVF and FLO.

To complement the factor analysis, we analyzed a correlation network based on Pearson's correlation (Figure 1). The analysis of the correlation network made it possible to identify the groupings of variables in the four factors similarly identified. Higher correlations were observed between the variables within the factors, and lower correlations were observed between variables belonging to different factors.

Figure 1

Correlation network between sensory quality (FIN = final score; ARO = fragrance/aroma; FLA = flavor; BOD = body; ACI = acidity; BAL = balance; OVE = overall; AFT = aftertaste), postharvest traits (UNR = unripe fruit; RIF = ripe fruit; SEM = semi-ripe fruit; OVF = overripe fruit; FLO = floating fruit), and yield (BRI = °Brix; YLD = yield; LBG = volume of cherry coffee per bag; DEF = defects; WEI = weight of 100 beans; S17 = sieve 17; S15 = sieve 15; B15 = bottom sieve; T19 = sieve 17 together with sieve 19) in Coffea arabica. Red lines represent negative correlations, and green lines represent positive correlations. The thickness of the line is proportional to the magnitude of the correlation. The highlighted lines show a correlation greater than 0.5 in modulus.

Our results show that the most substantial gains were achieved in traits related to sieve quality (B15, S17, S15, and T19) and fruit ripening uniformity (RIF, UNR, OVF, SEM, and FLO) (Table 4). In contrast, gains in sensory attributes, although present, were more modest, ranging between 1 % and 2 %. Genetic gains for yield were about 14%. These findings highlight the effectiveness of both direct selection and multivariate genetic gain through factor analysis in enhancing various coffee traits.

Thumbnail

Table 4
Direct gain and multivariate gain based on the sensory quality factor.

{\bar{X}}_{0}

is the mean population.

{\bar{X}}_{S}

is the mean of selected individuals based on the direct selection on the th variable.

{\bar{X}}_{SM}

is the mean of individuals selected based on the sensory quality factor. ΔG and ΔGM represent the direct gain and the multivariate gain, respectively.

Based on the selection made in the sensory quality factor, the gains obtained in the sensory attributes were maintained as regards the direct gain, as expected (Table 4). As for the sieves factor, the direction of the selection was altered. In other words, it is desirable to increase the bean size, but selection using the sensory quality factor can result in a reduction in bean size by reducing the percentage of S17 and T19. As regards the fruit ripening uniformity factor, we also observed that the UNR trait was selected in the opposite direction (thereby increasing the percentage of green coffee), which is not desirable for producing quality coffee. For the overripe fruit factor, there was a reduction in genetic gain, but it remained in the desired direction, with a reduction in OVF and FLO. Finally, yield (YLD) was negatively affected by selection based on the sensory quality factor.

Discussion

In this study, we investigated the genetic parameters and correlations in C. arabica for the main traits related to coffee quality, post-harvest characteristics, and productivity. In a population of 59 C. arabica progenies, we analyzed the most important quality-related traits. Our analysis revealed significant genetic variation in all traits. In addition, we examined the effectiveness of direct selection and multivariate genetic gain using factor analysis.

The estimated heritability and accuracy varied widely between variables, reflecting differences in genetic control and the potential for selection. Sensory attributes showed moderate to high heritability (19.5 % to 71.3 %), indicating considerable genetic variability that can be exploited in breeding programs. These findings align with the study by Cheserek et al. (2022), which reported significant genetic variability and high heritability (> 60 %) for sensory traits such as taste and aroma. However, our results also revealed lower heritability for specific sensory attributes, such as fragrance/aroma (19.47%), consistent with previous reports of limited genetic control over sensory traits (Malau et al., 2018). These differences confirm that the heritability of sensory quality is influenced by specific factors of the experiment, such as the populations studied or the evaluation methods. In contrast, other quality-related traits, such as bean size and ripening uniformity, showed high genotypic coefficients and heritability.

The predicted genetic gains for ripening uniformity and coffee sieve quality (about 25 %) and yield (about 14 %) were high. On the other hand, direct genetic gains for sensory traits were lower (1-2 %). This is to be expected given the complexity of coffee quality variation resulting from the interaction of multiple traits and environmental influences (Ferreira et al., 2005; Paixão et al., 2022). These results suggest an effective strategy for enhancing coffee quality by targeting indirect traits rather than focusing solely on sensory quality. Therefore, it is essential to consider the relationships between various quality variables and other agronomic traits to establish more effective selection strategies.

As regards the relationship between variables, our results showed low Pearson correlation between sensory attributes and other quality-related traits, such as bean size, ripening uniformity, and yield. Although an increase in bean size generally indicates superior coffee quality (Lima et al., 2020; Velásquez and Banchón, 2023), our results revealed weak correlations between the sensory attributes and sieve sizes (S15, S17, and P19). The low correlation found between sensory attributes and other quality-related characteristics can be attributed to the processing of samples for sensory analysis according to the SCAA protocol, which removes defective beans and standardizes the samples for analysis. Although this practice is essential to ensure the accuracy of sensory analysis, it can obscure biological connections between variables.

Factor analysis effectively grouped the correlated characteristics into four latent variables: sensory quality, sieves (bean size), ripening uniformity, and overripe fruit. Obtaining interpretable factors shows that the variables evaluated have a correlation pattern that can be summarized using common factors. These factors make sense from an agronomic point of view when discussing coffee breeding (Ferreira et al., 2005; Barbosa al., 2019; Paixão et al., 2022).

The use of factorial scores as selection criteria can optimize the selection process by preserving the biological and agronomic significance of the correlated traits (Ferreira et al., 2005; Barbosa et al., 2019; Piza et al., 2023). The sensory quality factor captures the combined effects such as aroma, flavor, acidity, body, and balance, allowing for the simultaneous selection of overall cup quality. This internal consistency highlights the potential for optimizing genetic gains for sensory quality by evaluating the final score rather than focusing on individual sensory components. This approach is particularly beneficial due to the high heritability of the final score, which was observed to be 71.25 %.

Selection based on the sieve factor enables breeders to select the ideal bean size distribution, which directly impacts grading and market price. The ripening uniformity and overripe fruit factors allow for simultaneous improvement in harvesting efficiency and a reduction in undesirable post-harvest defects. This multivariate selection strategy is particularly advantageous for perennial crops, such as coffee, where the breeding cycle is long and genetic progress can be slow (Piza et al., 2023). By selecting several correlated traits simultaneously, breeders can accelerate the development of new cultivars with enhanced sensory quality, productivity, and post-harvest attributes (Barbosa et al., 2019).

One of the important aspects of multivariate selection is to check for correlated responses in other characteristics. We tested this by carrying out a direct selection based on the sensory quality factor. The results suggest that focusing exclusively on sensory quality can result in a reduction in bean size and overall yield. The low correlation observed between sensory attributes and other agronomic traits suggests that selection strategies should be balanced to avoid undesirable responses. To optimize selection gains, we recommend direct selection for highly heritable traits such as bean size and fruit ripening uniformity. The use of factor analysis in selection for quality can serve as a strategy to monitor simultaneous genetic gains without compromising essential agronomic traits such as yield.

Declaration of use of AI Technologies
We used Elicit^® and Scispace^® for literature review, and Grammarly^® and ChatGPT^® for English language revision.

Acknowledgments

The authors would like to thank the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES), Finance code 001, (grant 88887.499809-2020-00) and Fundação de Amparo à Pesquisa do Estado de Minas Gerais (FAPEMIG) (grant 1568 - 655) for Masters scholarships; and Coffee Research Consortium (grant 10.18.20.037.00.00) for supporting this study.

Data availability statement

All datasets were generated and analyzed in the current study.

References

Barbosa IP, Costa WG, Nascimento M, Cruz CD, Oliveira ACB. 2019. Recommendation of Coffea arabica genotypes by factor analysis. Euphytica 215: 1-10. https://doi.org/10.1007/s10681-019-2499-x
» https://doi.org/10.1007/s10681-019-2499-x
Cheserek JJ, Ngugi K, Muthomi JW, Omondi CO, Kathurima CW. 2022. Genetic variability and correlation of biochemical and sensory characteristics of coffee. Journal of Agricultural Science 14: 95-103. https://doi.org/10.5539/jas.v14n2p95
» https://doi.org/10.5539/jas.v14n2p95
Cruz CD, Carneiro PCS, Regazzi AJ. 2012 Modelos Biométricos Aplicados ao Melhoramento Genético. Editora UFV, Viçosa, MG, Brazil (in Portuguese).
Cruz CD. 2013. GENES - a software package for analysis in experimental statistics and quantitative genetics. Acta Scientiarum 35: 271-276. https://doi.org/10.4025/actasciagron.v35i3.21251
» https://doi.org/10.4025/actasciagron.v35i3.21251
Dessalegn Y, Labuschagne MT, Osthoff G, Herselman L. 2008. Genetic diversity and correlation of bean caffeine content with cup quality and green bean physical characteristics in coffee (Coffea arabica L.). Journal of the Science of Food and Agriculture 88: 1726-1730. https://doi.org/10.1002/jsfa.3271
» https://doi.org/10.1002/jsfa.3271
Epskamp S, Cramer AOJ, Waldorp LJ, Schmittmann VD, Borsboom D. 2012. qgraph: network visualizations of relationships in psychometric data. Journal of Statistical Software 48: 1-18. https://doi.org/10.18637/jss.v048.i04
» https://doi.org/10.18637/jss.v048.i04
Ferreira A, Cecon PR, Cruz CD, Ferrão RG, Silva MF, Fonseca AFA, et al. 2005. Simultaneous selection of Coffea canephora by means of combination of factor analysis and selection indexes. Pesquisa Agropecuária Brasileira 40: 1189-1195. https://doi.org/10.1590/S0100-204X2005001200005
» https://doi.org/10.1590/S0100-204X2005001200005
González AL, Lopez AM, Gaytán ORT, Ramos VM. 2019. Cup quality attributes of Catimors as affected by size and shape of coffee bean (Coffea arabica L.). International Journal of Food Properties 22: 758-767. https://doi.org/10.1080/10942912.2019.1603997
» https://doi.org/10.1080/10942912.2019.1603997
Granate MJ, Cruz CD, Cecon PR, Pacheco CAP. 2001. Analysis of prediction factors in yields by maize selection (Zea mays L.). Acta Scientiarum 23: 1271-1279 (in Portuguese, with abstract in English).
Kathurima CW, Gichimu BM, Kenji GM, Muhoho SM, Boulanger R. 2009. Evaluation of beverage quality and green bean physical characteristics of selected Arabica coffee genotypes in Kenya. African Journal of Food Science 3: 365-371.
Lima JSS, Silva SA, Fonseca A, Pajehu LF. 2020. Quality of Coffea canephora beverage as a function of genotype, processing method and grain size. Coffee Science 15: e151714. https://doi.org/10.25186/.v15i.1714
» https://doi.org/10.25186/.v15i.1714
Malau S, Siagian A, Sirait B, Ambarita H, Pandiangan S, Sihotang MR, et al. 2018. Variability of organoleptic quality of Arabica coffee. Anatolian Journal of Agricultural Sciences 33: 241-245 (in Turkish, with abstract in English). http://doi.org/10.7161/omuanajas.405418
» http://doi.org/10.7161/omuanajas.405418
Montagnon C, Guyot B, Cilas C, Leroy T. 1998. Genetics parameters of several biochemical compounds from green coffee, Coffea canephora Plant Breeding 117: 576-578. https://doi.org/10.1111/j.1439-0523.1998.tb02211.x
» https://doi.org/10.1111/j.1439-0523.1998.tb02211.x
Paixão PTM, Nascimento ACC, Nascimento M, Azevedo CF, Oliveira GF, Silva FL, et al. 2022. Factor analysis applied in genomic selection studies in the breeding of Coffea canephora Euphytica 218: 42. https://doi.org/10.1007/s10681-022-02998-x
» https://doi.org/10.1007/s10681-022-02998-x
Piza MR, Luz SROT, Andrade VT, Figueiredo VC, Abrahão JCR, Bruzi AT, et al. 2023. Multiple traits selection strategies: a proposal for coffee plant breeding. Agronomy 13: 2033. https://doi.org/10.3390/agronomy13082033
» https://doi.org/10.3390/agronomy13082033
Reyes González F, Prado EE, Portilla EP, Vargas GA, Curiel-Rodríguez AC, Gómez JAH. 2016. Evaluation of productivity, physical and sensory quality of coffee (Coffea arabica L.), beans in coffee trees grafted on CRUO, Huatusco, Veracruz. Revista de Geografía Agrícola 56: 45-53 (in Spanish, with abstract in English). https://doi.org/10.5154/r.rga.2016.56.006
» https://doi.org/10.5154/r.rga.2016.56.006
Rosado RDS, Rosado LDS, Cremasco JPG, Santos CEM, Dias DCFS, Cruz CD. 2017. Genetic divergence between passion fruit hybrids and reciprocals based on seedling emergence and vigor. Journal of Seed Science 39: 417-425. https://doi.org/10.1590/2317-1545v39n4183293
» https://doi.org/10.1590/2317-1545v39n4183293
Rosado RDS, Rosado LDS, Borges LL, Bruckner CH, Cruz CD, Santos CEM. 2019. Genetic diversity of sour passion fruit revealed by predicted genetic values. Agronomy Journal 111: 165-174. https://doi.org/10.2134/agronj2018.05.0310
» https://doi.org/10.2134/agronj2018.05.0310
Sobreira FM, Oliveira ACB, Pereira AA, Guarçoni AM, Sakiyama NS. 2016. Divergence among Arabica coffee genotypes for sensory quality. Australian Journal of Crop Science 10: 1442-1448. http://doi.org/10.21475/ajcs.2016.10.10.p7430
» http://doi.org/10.21475/ajcs.2016.10.10.p7430
Specialty Coffee Association of America (SCAA). 2015. SCAA Protocols: Cupping Specialty Coffee. SCAA, Santa Ana, CA, USA. Available at http://www.scaa.org/?page=resources&d=cupping-protocols [Accessed Jan 20, 2024]
» http://www.scaa.org/?page=resources&d=cupping-protocols
Velásquez S, Banchón C. 2023. Influence of pre-and post-harvest factors on the organoleptic and physicochemical quality of coffee: a short review. Journal of Food Science and Technology 60: 2526-2538. https://doi.org/10.1007/s13197-022-05569-z
» https://doi.org/10.1007/s13197-022-05569-z

Edited by

Edited by:
Leonardo Oliveira Medici

Publication Dates

Publication in this collection
21 Nov 2025
Date of issue
2025

History

Received
15 Jan 2025
Accepted
13 Mar 2025

This is an Open Access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

[1] Barbosa IP, Costa WG, Nascimento M, Cruz CD, Oliveira ACB. 2019. Recommendation of Coffea arabica genotypes by factor analysis. Euphytica 215: 1-10. https://doi.org/10.1007/s10681-019-2499-x
» https://doi.org/10.1007/s10681-019-2499-x

[2] Cheserek JJ, Ngugi K, Muthomi JW, Omondi CO, Kathurima CW. 2022. Genetic variability and correlation of biochemical and sensory characteristics of coffee. Journal of Agricultural Science 14: 95-103. https://doi.org/10.5539/jas.v14n2p95
» https://doi.org/10.5539/jas.v14n2p95

[3] Cruz CD, Carneiro PCS, Regazzi AJ. 2012 Modelos Biométricos Aplicados ao Melhoramento Genético. Editora UFV, Viçosa, MG, Brazil (in Portuguese).

[4] Cruz CD. 2013. GENES - a software package for analysis in experimental statistics and quantitative genetics. Acta Scientiarum 35: 271-276. https://doi.org/10.4025/actasciagron.v35i3.21251
» https://doi.org/10.4025/actasciagron.v35i3.21251

[5] Dessalegn Y, Labuschagne MT, Osthoff G, Herselman L. 2008. Genetic diversity and correlation of bean caffeine content with cup quality and green bean physical characteristics in coffee (Coffea arabica L.). Journal of the Science of Food and Agriculture 88: 1726-1730. https://doi.org/10.1002/jsfa.3271
» https://doi.org/10.1002/jsfa.3271

[6] Epskamp S, Cramer AOJ, Waldorp LJ, Schmittmann VD, Borsboom D. 2012. qgraph: network visualizations of relationships in psychometric data. Journal of Statistical Software 48: 1-18. https://doi.org/10.18637/jss.v048.i04
» https://doi.org/10.18637/jss.v048.i04

[7] Ferreira A, Cecon PR, Cruz CD, Ferrão RG, Silva MF, Fonseca AFA, et al. 2005. Simultaneous selection of Coffea canephora by means of combination of factor analysis and selection indexes. Pesquisa Agropecuária Brasileira 40: 1189-1195. https://doi.org/10.1590/S0100-204X2005001200005
» https://doi.org/10.1590/S0100-204X2005001200005

[8] González AL, Lopez AM, Gaytán ORT, Ramos VM. 2019. Cup quality attributes of Catimors as affected by size and shape of coffee bean (Coffea arabica L.). International Journal of Food Properties 22: 758-767. https://doi.org/10.1080/10942912.2019.1603997
» https://doi.org/10.1080/10942912.2019.1603997

[9] Granate MJ, Cruz CD, Cecon PR, Pacheco CAP. 2001. Analysis of prediction factors in yields by maize selection (Zea mays L.). Acta Scientiarum 23: 1271-1279 (in Portuguese, with abstract in English).

[10] Kathurima CW, Gichimu BM, Kenji GM, Muhoho SM, Boulanger R. 2009. Evaluation of beverage quality and green bean physical characteristics of selected Arabica coffee genotypes in Kenya. African Journal of Food Science 3: 365-371.

[11] Lima JSS, Silva SA, Fonseca A, Pajehu LF. 2020. Quality of Coffea canephora beverage as a function of genotype, processing method and grain size. Coffee Science 15: e151714. https://doi.org/10.25186/.v15i.1714
» https://doi.org/10.25186/.v15i.1714

[12] Malau S, Siagian A, Sirait B, Ambarita H, Pandiangan S, Sihotang MR, et al. 2018. Variability of organoleptic quality of Arabica coffee. Anatolian Journal of Agricultural Sciences 33: 241-245 (in Turkish, with abstract in English). http://doi.org/10.7161/omuanajas.405418
» http://doi.org/10.7161/omuanajas.405418

[13] Montagnon C, Guyot B, Cilas C, Leroy T. 1998. Genetics parameters of several biochemical compounds from green coffee, Coffea canephora Plant Breeding 117: 576-578. https://doi.org/10.1111/j.1439-0523.1998.tb02211.x
» https://doi.org/10.1111/j.1439-0523.1998.tb02211.x

[14] Paixão PTM, Nascimento ACC, Nascimento M, Azevedo CF, Oliveira GF, Silva FL, et al. 2022. Factor analysis applied in genomic selection studies in the breeding of Coffea canephora Euphytica 218: 42. https://doi.org/10.1007/s10681-022-02998-x
» https://doi.org/10.1007/s10681-022-02998-x

[15] Piza MR, Luz SROT, Andrade VT, Figueiredo VC, Abrahão JCR, Bruzi AT, et al. 2023. Multiple traits selection strategies: a proposal for coffee plant breeding. Agronomy 13: 2033. https://doi.org/10.3390/agronomy13082033
» https://doi.org/10.3390/agronomy13082033

[16] Reyes González F, Prado EE, Portilla EP, Vargas GA, Curiel-Rodríguez AC, Gómez JAH. 2016. Evaluation of productivity, physical and sensory quality of coffee (Coffea arabica L.), beans in coffee trees grafted on CRUO, Huatusco, Veracruz. Revista de Geografía Agrícola 56: 45-53 (in Spanish, with abstract in English). https://doi.org/10.5154/r.rga.2016.56.006
» https://doi.org/10.5154/r.rga.2016.56.006

[17] Rosado RDS, Rosado LDS, Cremasco JPG, Santos CEM, Dias DCFS, Cruz CD. 2017. Genetic divergence between passion fruit hybrids and reciprocals based on seedling emergence and vigor. Journal of Seed Science 39: 417-425. https://doi.org/10.1590/2317-1545v39n4183293
» https://doi.org/10.1590/2317-1545v39n4183293

[18] Rosado RDS, Rosado LDS, Borges LL, Bruckner CH, Cruz CD, Santos CEM. 2019. Genetic diversity of sour passion fruit revealed by predicted genetic values. Agronomy Journal 111: 165-174. https://doi.org/10.2134/agronj2018.05.0310
» https://doi.org/10.2134/agronj2018.05.0310

[19] Sobreira FM, Oliveira ACB, Pereira AA, Guarçoni AM, Sakiyama NS. 2016. Divergence among Arabica coffee genotypes for sensory quality. Australian Journal of Crop Science 10: 1442-1448. http://doi.org/10.21475/ajcs.2016.10.10.p7430
» http://doi.org/10.21475/ajcs.2016.10.10.p7430

[20] Specialty Coffee Association of America (SCAA). 2015. SCAA Protocols: Cupping Specialty Coffee. SCAA, Santa Ana, CA, USA. Available at http://www.scaa.org/?page=resources&d=cupping-protocols [Accessed Jan 20, 2024]
» http://www.scaa.org/?page=resources&d=cupping-protocols

[21] Velásquez S, Banchón C. 2023. Influence of pre-and post-harvest factors on the organoleptic and physicochemical quality of coffee: a short review. Journal of Food Science and Technology 60: 2526-2538. https://doi.org/10.1007/s13197-022-05569-z
» https://doi.org/10.1007/s13197-022-05569-z

Access code	Identification	Description	Origin 1	Origin 2
UFV-CRP 01	Arara	Late maturation, São Paulo Ranch, great material, SSP, Sarchimor FL-Introgressed	IAC 1669-20-F2	Icatu 2944
UFV-CRP 02	Sabiá 398	99-8, great material, late maturation	Catimor UFV 386	Acaia 474-19
UFV-CRP 03	Icatu 3696	Cv 525 3.21, Tuiuiu, Icatu derived- Introgressed	Coffea canephora	Coffea arabica
UFV-CRP 04	Siriema 14/8	Rio Fundo, CK, Racemosa derived- Introgressed	Coffea racemosa	Coffea arabica (cv. Blue Mountain)
UFV-CRP 05	Catucaí Amarelo	3 SM Rio Fundo, MF, CK, Icatu derived- Introgressed	Icatu Amarelo	Catuaí Amarelo
UFV-CRP 06	Siriema 5/14	3 FSA, Racemosa derived- Introgressed	Coffea racemosa	Coffea arabica (cv. Blue Mountain)
UFV-CRP 07	Catucaí Vermelho 36/6	Progeny 366 (3.12), good, Icatu derived- Introgressed	Icatu Vermelho	Catuaí Vermelho
UFV-CRP 08	Catucaí Amarelo	Bronze color vegetative buds, FSA, hybrid, Icatu derived- Introgressed	Icatu Amarelo	Catuaí Amarelo
UFV-CRP 09	Maracatia	Low level of the trunk (Red), Other fixed lines-Typica/Bourbon	Acaiá IAC 474	Catucaí Vermelho IAC 81
UFV-CRP 10	Maracatia	Low level of the trunk (yellow), Other fixed lines-Typica/Bourbon	Acaiá IAC 474	Catucaí Vermelho IAC 81
UFV-CRP 11	Acauã 65	Origin 37 Abacateiro, FSA, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 12	Catucaí Vermelho 20/15	Medium size, high vigor, FSA, very good material, Icatu derived- Introgressed	Icatu Vermelho	Catuaí Vermelho
UFV-CRP 13	Acauã 65	Great material, Icatu derived- Introgressed	Sarchimor 1668	Mundo Novo IAC 388-17
UFV-CRP 14	Acauã 68/11	Cv 48 (Item 8 MG 3.45), good material, Icatu derived- Introgressed	Sarchimor 1668	Mundo Novo IAC 388-17
UFV-CRP 15	Icatu Amarelo 925	F2, TA (SSP 0,25), F3, segregation (fruit color), hybrid, Icatu derived-Introgressed	Coffea canephora	Coffea arabica
UFV-CRP 16	Icatu Vermelho 925	F2, TA (SSP 0,25), F3, segregation (fruit color), hybrid, Icatu derived-Introgressed	Coffea canephora	Coffea arabica
UFV-CRP 17	Catucaí Amarelo 2 SL	Procafé Fundation, good material, segregation, mediun size, Icatu derived- Introgressed	Icatu Amarelo	Catuaí Amarelo
UFV-CRP 18	Catuaí Vermelho 19/8	Cv 380Japi	IAC 476-11	Mundo Novo IAC 374-19
UFV-CRP 19	Catucaí -Açu	Ranch Santa Maria, CK, 83, good material	Icatu Vermelho	Catuaí Vermelho
UFV-CRP 20	Águia	Great material	Catuaí Vermelho (?)	Catimor 357-77 (5/33)
UFV-CRP 21	Catucaí Amarelo 24/137	Bronze bud, Germplasm Bank of M. Soares, Icatu derived- Introgressed	Icatu Amarelo	Catuaí Amarelo
UFV-CRP 22	Acauã Laranja	F3 hybrid, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 23	Acauã Amarelo	F3 hybrid, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 24	Acauã Vermelho	F3 hybrid, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 25	Acauã	D. Martins V SSP 4 %, great material, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 26	Catucaí Amarelo 24/137	Green color vegetative buds, Germplasm Bank of M. Soares, Icatu derived- Introgressed	Icatu Amarelo	Catuaí Amarelo
UFV-CRP 27	Catucaí Vermelho 20/15	Santa Maria Ranch, CK, good material, Icatu derived- Introgressed	Icatu Vermelho	Catuaí Vermelho
UFV-CRP 28	Acauã	D. Martins II 7 %, F3 hybrid, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 29	Palma III	Santo Antônio Farm	IAC 81	Catimor UFV 353
UFV-CRP 30	Catucaí Amarelo	CK, Rio Fundo, good vigor, Icatu derived- Introgressed	Icatu Amarelo	Catuaí Amarelo
UFV-CRP 31	Acauã 68/2	Cv 9 (Item 2 MG 3.45), good material, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 32	Bem Te Vi	Cv 334 (3.27), good material	Catimor 391	Catuai Amarelo IAC 74
UFV-CRP 33	Icatu 925	Cv 231 (3.25), Clone Carlos, Icatu derived- Introgressed	Coffea canephora	Coffea arabica
UFV-CRP 34	Siriema Vermelho	Resistant material (CEPEC), Racemosa derived-Introgressed	Coffea racemosa	Coffea arabica (cv. Blue Mountain)
UFV-CRP 35	Siriema Amarelo	Resistant material (CEPEC), Racemosa derived-Introgressed	Coffea racemosa	Coffea arabica (cv. Blue Mountain)
UFV-CRP 36	Catucaí 785/15	Jaguaraí, , Icatu derived- Introgressed	Icatu Vermelho 785	Catuaí Vermelho
UFV-CRP 37	Catucaí Amarelo 3/5	Cv 749 (3-27), big fruits production, Icatu derived- Introgressed	Icatu Amarelo	Catuaí Amarelo
UFV-CRP 38	Catucaí Amarelo 3/5	Cv 749 (3-27), big fruits production, LVA, Icatu derived- Introgressed	Icatu Amarelo	Catuaí Amarelo
UFV-CRP 39	Catucaí Amarelo 24/137	Bronze color vegetative buds, Germplasm Bank of M. Soares, Icatu derived- Introgressed	Icatu Amarelo	Catuaí Amarelo
UFV-CRP 40	Catucaí Vermelho 20/15	Second selection, FSA, Icatu derived- Introgressed	Icatu Vermelho	Catuaí Vermelho
UFV-CRP 41	Acauã	D. Martins SSP 3 %, great coffee, F3, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 42	Acauã	D. Martins SSP 5 %, good material, F3, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 43	Catucaí Vermelho	Cv 446 (3.27), Icatu derived- Introgressed	Icatu Vermelho	Catuaí Vermelho
UFV-CRP 44	Acauã Amarelo	D. Martins II SSP 10 %, late maturation, good material, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 45	Acauã Amarelo	D. Martins II SSP 10 %, medium ripeness, good material, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 46	Araponga	EPAMIG, Cavimor FL (2570)-Introgressed	IAC 86	Híbrido de Timor UFV 446/08
UFV-CRP 47	Acauã Amarelo	D. Martins I, good material, Sarchimor FL-Introgressed	Sarchimor 1668	Mundo Novo IAC 388-17
UFV-CRP 48	Icatu 925	SSP TA, hybrid, VAL, Icatu derived- Introgressed	Coffea canephora	Coffea arabica
UFV-CRP 49	Acauã	Cv 106 (Item 18 3.45), Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 51	Acauã Amarelo	F2, São Paulo Ranch = Tião Andrade, hybrid, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 52	Acauã Vermelho	F2, São Paulo Ranch = Tião Andrade, hybrid, Sarchimor FL-Introgressed	Sarchimor IAC 1668	Mundo Novo IAC 388-17
UFV-CRP 53	IPR 103	Iapar, Icatu derived-Introgressed	Catuaí	Icatu
UFV-CRP 54	Catucaí Vermelho	Rio Fundo CK, JAPI, Icatu derived- Introgressed	Icatu Vermelho	Catuaí Vermelho
UFV-CRP 55	3 SM	RII, Cv 15	Unknown	Unknown
UFV-CRP 56	Catucaí Amarelo 20/15	(5-13), FSA, Icatu derived- Introgressed	Icatu Amarelo	Catuaí Amarelo
UFV-CRP 57	Catucaí Vermelho 36/69	(5/42), FSA W366, Icatu derived- Introgressed	Icatu Vermelho	Catuaí Vermelho
UFV-CRP 58	Acauã Amarelo	FSA, segregation (level of the trunk), Sarchimor FL-Introgressed	Sarchimor 1668	Mundo Novo IAC 388-17
UFV-CRP 59	Catucaí Roxinho	Cv 346 (3-27), Icatu derived- Introgressed	Icatu Vermelho	Catuaí Vermelho
UFV-CRP 60	IPR 98	Iapar, Sarchimor FL-Introgressed	Villa Sarchi CIFC 971/10	Híbrido de Timor CIFC 832/2

Variable	F-value	p-value	Heritability	Accuracy	${\hat{σ}}_{p}^{2}$	${\hat{σ}}_{e}^{2}$	${\hat{σ}}_{g}^{2}$	CV_g	$C V_{g} C V_{e}^{- 1}$
			---------------- % ----------------					%
FIN	3.478	0.000**	71.25	84.41	1.511	0.434	1.077	12.680	0.90
ARO	1.241	0.162ns	19.47	44.12	0.013	0.011	0.002	0.688	0.28
FLA	2.755	0.000**	63.72	79.82	0.044	0.016	0.028	2.254	0.76
ACI	3.316	0.000**	69.85	83.57	0.038	0.011	0.026	2.215	0.87
BOD	3.001	0.000**	66.69	81.66	0.036	0.012	0.024	2.093	0.81
OVE	2.213	0.002**	54.83	74.05	0.045	0.020	0.024	2.133	0.63
BAL	3.290	0.000**	69.61	83.43	0.042	0.012	0.029	2.335	0.87
AFT	3.150	0.000**	68.26	82.62	0.040	0.012	0.027	2.272	0.84
B15	2.862	0.000**	65.06	80.66	15.5656	5.438	10.126	29.323	0.78
S15	3.562	0.000**	71.93	84.81	83.7724	23.516	60.256	19.004	0.92
S17	3.399	0.000**	70.58	84.02	118.183	34.761	83.421	18.998	0.89
T19	3.399	0.000**	70.59	84.02	118.183	34.761	83.421	18.998	0.89
YLD	2.241	0.000**	55.34	74.43	53.372	23.808	29.56	11.34	0.64
LBG	1.905	0.002**	47.52	68.94	197.069	103.42	93.64	3.55	0.54
WEI	2.804	0.000**	64.35	80.22	0.648	0.231	0.417	4.58	0.77
DEF	2.605	0.000**	61.62	78.50	7.157	2.747	4.410	23.13	0.73
BRI	1.829	0.003**	45.35	67.34	2.331	1.274	1.057	5.49	0.52
RIF	1.974	0.001**	49.35	70.25	0.009	0.004	0.004	11.93	0.56
UNR	2.861	0.000**	65.05	80.65	0.006	0.002	0.004	41.74	0.78
SEM	1.721	0.007**	41.91	64.74	0.005	0.002	0.002	21.50	0.49
OVF	1.654	0.011*	39.56	62.89	0.001	0.0007	0.0005	34.98	0.46
FLO	2.160	0.000**	53.72	73.29	0.001	0.0007	0.0008	22.94	0.62

Variable	Final factor loadings
Variable	Factor I	Factor II	Factor III	Factor IV	C
Sensory quality factor
FIN	0.995				0.993
ARO	0.758				0.579
FLA	0.963				0.931
ACI	0.933				0.893
BOD	0.910				0.835
OVE	0.931				0.890
BAL	0.939				0.886
AFT	0.917				0.844
Sieves factor
B15		–0.548			0.539
S15		–0.903			0.863
S17		0.972			0.974
T19		0.972			0.974
WEI		0.609			0.388
Fruit ripening uniformity factor
RIF			–0.738		0.660
UNR			0.807		0.702
LBG			_–0.728		0.548
BRI			–0.698		0.635
Overripe fruit factor
YLD				_–0.591	0.370
OVF				0.857	0.765
SEM				_–0.604	0.555
FLO				0.564	0.632
DEF				0.239	0.239

Variable	${\bar{X}}_{0}$	${\bar{X}}_{S}$	ΔG(%)	${\bar{X}}_{SM}$	ΔG_M(%)
Sensory quality factor
FIN	81.84	83.81	1.72	83.81	1.72
ARO	7.52	7.70	0.49	7.65	0.34
FLA	7.46	7.79	2.79	7.79	2.79
ACI	7.40	7.71	2.92	7.69	2.74
BOD	7.42	7.73	2.76	7.70	2.51
OVE	7.37	7.73	2.70	7.71	2.56
BAL	7.36	7.69	3.09	7.68	3.00
AFT	7.31	7.60	2.72	7.59	2.63
Sieves factor
B15	10.85	6.28	–27.38	11.46	3.65
S15	40.85	26.52	–25.22	43.48	4.63
S17	48.07	65.41	25.45	44.61	–5.09
T19	48.07	65.41	25.45	44.61	–5.09
WEI	14.11	15.38	5.76	13.85	–1.21
Fruit ripening uniformity factor
RIF	0.568	0.71	12.52	0.589	1.84
UNR	0.153	0.06	–38.62	0.179	11.39
LBG	272.21	292.83	3.60	269.65	–0.45
BRI	18.74	21.23	6.03	18.48	–0.63
Overripe fruit factor
YLD	47.95	59.85	13.75	44.41	–4.09
OVF	0.064	0.02	–26.07	0.051	–7.92
SEM	0.214	0.11	–19.50	0.179	–6.82
FLO	0.129	0.08	–21.01	0.097	–12.99

Multi-trait selection and the relationship between sensory analysis and post-harvest variables in Coffea arabica

ABSTRACT

Introduction

Materials and Methods

Genetic material and phenotypic data

Statistical analysis

Results

Discussion

Acknowledgments

Data availability statement

References

Edited by

Publication Dates

History