2. Results and Discussion
Coffee cultivation in the Western Amazon region takes place in Am and Aw climates, which are characterized as tropical, warm, and humid [
12]. The total annual precipitation for the 2020/2021 crop year, spanning from July 2020 to July 2021, was 2.112 mm, while for the 2021/2022 crop year (July 2021 to July 2022), it was 1.847 mm (
Figure 1).
It is observed that the first measurement recorded higher annual accumulated precipitation, and with the exception of April 2022, the first crop also showed higher monthly accumulated precipitation (
Figure 1). It is known that water deficits in coffee plants can lead to flower drop and damage to fruit growth, resulting in reduced bean filling [
13,
14,
15]. Comparing the overall averages of the two measurements (
Table 1), it is noted that the parameters Total Titratable Acidity (TTA), Total Crude Protein (TCP), Total Soluble Solids (TSS), ratio, Soluble Sugars (SS), and Total Reducing Sugars (TRS) showed significant differences between the measurements, which may be influenced by environmental factors. However, Aqueous Extract (AE), Total Ash (TA), Hydrogen Potential (pH), Ether Extract (EE), Total Phenolic Compounds (TPC), and Non-Reducing Sugars (NRS) did not show significant differences between the measurements.
The average concentrations of the variables TCP, TSS, ratio, SS, and TRS were higher in the first measurement, where the annual accumulated precipitation was greater. In contrast, the average TTA content was higher in the second measurement (
Table 1).
Genotype × Years (G × Y) interactions were significant for all physicochemical traits, indicating the presence of genotypes with different performance between measurements (
Table 1). The analysis of repeated measures over time allows us to consider temporary and permanent effects of experimental error. The genotype × years interaction was classified as simple (
Figure 2), considering that changes in performance resulted in only small changes in the genotypic classification.
Repeatability, also referred to as the upper limit of heritability, is estimated by considering both its permanent and temporary components. Since this estimate takes into account not only additive genetic variance but also environmental factors that consistently impact the individual over time. Except for pH, which had a low repeatability estimate (r = 34.91), the other traits had repeatability estimates ranging from 55.93 to 74.42 (
Table 1), which can be interpreted as moderate to high magnitudes [
16]. The traits can be ranked based on their repeatability estimates as follows: TCP > AE ≥ EE > TPC > TA > TTA > SS ≥ NRS > TRS > TSS > ratio > pH.
The experimental coefficient of variation (CVe), estimated based on the average performance of the traits and the experimental error estimate, was assessed to gauge the precision of the experiments. All traits showed low CVe estimates, indicating good experimental precision, with values ranging from 0.37% for non-reducing sugars to 6.41% for pH (
Table 1). Ratios greater than one between the genotypic coefficient of variation (CVg) and the coefficient of variation (CVe) indicate genetic progress through plant selection [
17]. This ratio ranged from 0.92 for TSS to 4.81 for TPC. Based on this analysis, the traits can be ranked as follows: TPC > AE > TA > TCP > TRS > SS > NRS > EE > TTA > pH > Ratio > TSS.
The clustering of genotypes using Scott-Knott’s mean test indicates that some traits, such as Total Phenolic Compounds (TPC), exhibited high variability, forming 17 groups, while others, such as Total Soluble Solids (TSS), showed less variability, forming 4 groups. In our study, the levels of soluble sugars, proteins, and phenolic compounds varied from to 3.66 to 9,84%, 11.63 to 18.93%, and 4.2 to 5.86 g of gallic acid equivalent per 100 g, respectively (
Supplementary Materials). Reducing and non-reducing sugars varied from 0.86 to 1.97% and 2.23 to 8.05%, respectively. Among the non-reducing sugars, sucrose is the primary constituent of this class [
18], and higher levels of soluble sugars positively influence sucrose concentration.
The aqueous extract (AE) reflects the content of substances in coffee beans that are soluble in boiling water, while Total Soluble Solids (TSS) represent the compounds soluble in water at room temperature [
19]. These traits showed substantial genetic variability, with some clones, such as AS2, having high levels, while others, including GJ8, N2, and GB7, exhibited much lower levels. The WE content ranged from 27.38 to 33.21%, and the TSS ranged from 28.97 to 35.88% (
Supplementary Materials).
The relationship between the soluble fraction, specifically total soluble solids and total titratable acidity, is associated with the perception of sweetness. When this ratio is unbalanced, it can create a sensation of the product being “diluted” or “too acidic”. In our study, approximately 70% of the genotypes had a ratio exceeding 0.2. This finding aligns with a previous study [
20], which noted that these range are not related with intrinsic sweetness perception. However this trait is associated with low acidity and high levels of soluble compounds found in the beans of this species. In this study, the acidity of the green beans ranged from 140.23 mL to 184.33 (NaOH 0.1 mol·L
−1 em 100 g sample). The pH, which tends to be influenced by the acidity of the beans, ranged from 5.06 to 5.32 (
Supplementary Materials).
The ether extract (EE) levels ranged from 3.45 to 7.89%, with some genotypes, such as GJ30, exhibiting high concentrations (7.89%), and others, such as BRS2357, showing lower concentrations (3.45%) (
Supplementary Materials). Reported EE levels ranging from 3.76 to 6.48% within the same species [
21]. The observed EE content was higher than at 10.90%. Other study that compared to Apoatã, Bukobensis, Laurentii, Guarani, and Conilon, observed that Conilon genotypes had lower values, close to 7.30% [
22].
Total ash content ranged from 4.18 to 5.47% (
Supplementary Materials). The presence of nitrogen in the ash of
C. canephora is usually higher than that of other nutrients, followed by potassium, calcium, phosphorus, sulfur, magnesium, iron, boron, manganese, copper, and zinc [
23].
Analyzed the physicochemical characteristics of seeds from
C. canephora genotypes of the Apoatã variety. Their study found an average ash content of 4.07%, ranging from 3.45 to 5.96%; soluble sugars with an average of 4.22%, varying between 3.61 to 5.03%; an average ether extract of 5.03%, with values ranging from 3.76 to 6.48%; and crude protein with an average of 17.75%, ranging from 22.88% to 14.93% [
21]. These results may be considered lower compared to those observed in this study.
Due to their metabolic and physiological origins, the chemical compounds in coffee beans may exhibit correlations with each other. Out of 45 possible phenotypic correlations between ten characteristics, 12 were significant (
Table 2). The aqueous extract content showed a positive and significant phenotypic correlation with ash content (r
pe = 0.36 **), acidity (r
pe = 0.33 **), total soluble solids (r
pe = 0.24 *), and phenolic compounds (r
pe = 0.42**). Ash content had a positive and significant correlation with acidity (r
pe = 0.42), phenolic compounds (r
pe = 0.29 *), and reducing sugars (r
pe = 0.43 **). Acidity displayed a significant negative correlation with pH (r = −0.39 **) and positive correlations with proteins (r
pe = 0.35 **), phenolic compounds (r
pe = 0.60 **), and reducing sugars (r
pe = 0.36 **). pH showed a significant negative correlation with protein content (r
pe = −0.25 *).
Among these phenotypic correlations, approximately 66% were significant in terms of genotypic correlation (
Table 2), with similar signs and close magnitudes. This indicates that genetic factors had a stronger influence on the association between traits compared to environmental factors, as phenotypic correlations are derived from measurements affected by both genetic and environmental factors [
24].
Unlike the phenotypic correlation estimates, the traits TTA × TSS (r
ge = 0.27
+), TCP × TPC (r
ge = 0.44
++), and TPC × TRS (r
ge = 0.48
++) showed significant correlations solely at the genotypic level (
Table 2). Although simple correlations between AE × TSS, TA × TPC, TTS × TPC, and pH × TCP indicated significant associations, the genotypic correlation estimates suggest that these associations are not due to genetic effects.
The significant association between TTA and AE indicates that the solubility of organic acids present in coffee beans is enhanced when they come into contact with boiling water. Similarly, there is a correlation between TPC and AE. Another important association found in this study was between TTA and TCP. Studying the amino acid profile in Robusta coffee, found high levels of glutamic and aspartic acids, confirming the direct proportional relationship between these variables [
25].
In the dispersion of the first two principal components, genotypes that are closer together are more similar across all evaluated physicochemical traits simultaneously (
Figure 3). The projection of variables onto this dispersion shows that, except for ether extract (EE), genotypes located in the right quadrants generally have higher average values for the traits assessed.
The ideal references for maximum and minimum performance were identified within this dispersion (
Figure 3). The genotypes BRS3193, AS1, and AS2 were positioned near the high-performance ideotype, whereas N2, GB7, R22, and WP6 were closer to the low-performance ideal. The BRS3193 cultivar exhibited high levels of TTA, TSS, TPC, SS, and NRS. The AS1 genotype was notable for its high SST content, and the AS2 genotype for its high AE content.
Among the various cultivars, BRS2357 displayed high levels of TCP and TSS, while BRS2299 had a notably high TSS content. Of the publicly available genotypes, GJ30 was distinguished by its high pH and EE, whereas CA1 was noted for its high TTA and low ratio. AS3 was prominent for its elevated levels of TSS, SS, and NRS, while GJ25 showed high contents of TSS, SS, and NRS along with a lower pH. AS7 excelled in both TTA and NRS, and AR106 had high levels of TSS and NRS with a lower pH. GJ8 was recognized for its high TRS and low EA, and GJ3 had the highest TA content. Among the clones from Embrapa’s active germplasm bank in RO, BAG22 and BAG38 stood out for their high TSS levels, while BAG19 was notable for its high SS, NRS, and elevated pH.
A bean with distinct characteristics, such as higher acidity, astringency, and elevated sugar concentration, can be obtained from the cultivar BRS3193, as well as from AS3 and GJ25. These genotypes produce beans with greater sweetness and high levels of water-soluble substances. In contrast, GJ8 has lower solubility in boiling water but features higher levels of reducing sugars, such as glucose and fructose, in its composition.
In addition to being responsible for the sweet flavor of the beverage, sugars are precursors of taste and aroma, reacting in various ways (fragmentation, caramelization, or interaction with amino acids) [
26,
27].
C. canephora tends to achieve higher scores in cup tastings, demonstrating a superior metabolic profile when its soluble constituent concentrations are elevated [
5,
28]. In this study, we compared the gains from selection across a range of characteristics, focusing on the primary trait of total sugar content.
A genotype suitable for cultivation should display a range of favorable traits. Selection can be based on a single key trait or by using a selection index that simultaneously evaluates multiple traits. In this study, selecting for the primary trait (SS) resulted in a total gain of 32.2% (
Table 3). This gain was lower compared to the gains achieved using other selection indices, which showed higher magnitude estimates.
The genotype × ideotype index, which measures the Euclidean distance between the studied genotypes and an ideal plant with maximum performance, yielded the highest estimated gain (SG = 58.86). The Smith & Hazel index, which combines linear traits, showed the second highest gain estimate (SG = 57.19). The Mulamba and Mock index, which aggregates the rankings of genotypes based on their genetic values for each trait, provided the third highest gain estimate (SG = 52.96) (
Table 3).
Among the genotypes selected for the primary trait, only genotype BAG19, based on the projection of variables onto the PCA dispersion, is not positioned in the right-hand quadrant. In this quadrant, the cultivar BRS3193 and the clones AS3, AS7, AR106, and LB88 were selected in one or more selection indices. This suggests that these accessions tend to exhibit not only sweetness in their green beans but also a range of other favorable physicochemical characteristics.
In the case of the cultivar BRS3193, this genotype is noted for its commercial beverage quality and less pronounced nuances [
4]. It had some desirable chemical characteristics, including a trigonelline content of 0.85%, chlorogenic acid of 5.48%, and caffeine content exceeding 2.70% [
5]. AS7 has an average sensory score above 82 points, while AR106 has a score close to 80 points [
29].
Among the other ranked genotypes, the cultivar BRS2299 and clone CA1 were selected by all selection indices for their high potential in physicochemical bean quality. They were followed by GJ30, BAG22, AS1, and BAG19, which were selected in two indices, and subsequently by GJ8, BRS3213, AS6, and BRS2357, which were selected in only one index.
Selecting genotypes based on the total sugar content (
Table 3) in green coffee beans tends to favor materials with desirable characteristics for the industry, such as high water solubility. However, the clone AS7 is an exception, as it maintained an average value of EA below the mean (
Table 4).
In general (
Table 4), the 19 genotypes selected using various strategies demonstrate potential for grain production with high levels of TRS and TCP. This promotes the formation of melanoidins and other precursors through the Maillard reaction during roasting. With the exception of BAG19 and LB88, all other genotypes exhibit high levels of TPC, regardless of the selection strategy employed.
The concentration of physicochemical compounds in green coffee beans leads to constituents that are crucial for the quality of the beverage after roasting [
30]. And the interactions between them are associated with the quality of the beverage [
31]. This includes compounds that are predominantly found in the soluble fraction, such as sugars, proteins, and phenolic compounds. These compounds not only act as precursors to flavor and aroma [
32,
33] but also contribute to the body of the Beverage characteristic highly valued in the soluble coffee industry.
Table 5 shows the classification of genotypes based on the combined concentration of these three important classes of chemical compounds. According to this classification, which uses Skott-Knott mean clustering, the materials in this study were divided into 5 groups, with concentrations ranging from 23.44% to 33.12%. Higher values are observed for BRS3193 and BRS2357, which are plants with high potential for producing substances that could enhance the value of coffee beans. In contrast, lower values are found in the materials BG180, SK41, WP6, AS10, BAG24, BRS3137, BRS2336, BAG23, N2, and GJ20.
In this study, we assessed the genetic variability of coffee genotypes grown in Western Amazonia, focusing on their physicochemical characteristics across two harvests. The results revealed significant genetic diversity among the materials, identifying a group of promising genotypes for breeding programs and enabling the selection of superior genotypes with desirable traits. Selecting genotypes based on the total sugar content in their green beans tends to favor the development of materials with desirable traits for the industry, such as good solubility in water. Genetic progress estimates indicate a strong potential for successfully selecting plants that exhibit a range of superior quality attributes, rather than focusing on just one trait. This includes high levels of sugars, proteins, lipids, and phenolic compounds in their green beans. Consequently, materials selected for one or more of these traits are highly promising and could be introduced into Rondônia plantations by coffee growers for further genetic improvement.