Geographic location

The investigation was carried out on a site located in Coatepec, Veracruz, between the coordinates 19° 28’ 20” north latitude and 96° 54’ 16” west longitude, at an average height of 1 250 meters above sea level. Its surface is 24.5 hectares and its type of soil is of volcanic origin, known as Andosol and consists of ashes with high allophone content that gives it lightness and smoothness, as well as the ability to fix nutrients, especially phosphorus (^{INEGI, 2004}). According to ^{INEGI (2009)}, the climate is semi-warm humid with rains throughout the year, with an annual rainfall range of 1 100 to 2 100 mm. Its vegetation is forest or jungle and its minimum and maximum temperature is 10 ºC and 35 ºC, respectively.

Description of the farm

The vegetation of the farm is mainly: Inga sp, Inga edulis, Acacia Farnesiana, Leucaena leucocephala, Grevillea robusta and Bursera simaruba, which serve as shade for coffee. Also, there are guavas, oranges, banana trees and mangoes, among others. To a lesser extent, there are coyolera palms and some specimens of oak. The cultivated coffee varieties are high-bearing Arabic: Typica in 80% and Mundo Novo 20%. Its planting densities are 1 111 and 2 415 plants per ha, respectively. The first variety is planted at a distance of 3 x 3 m with an arrangement in three bolillo where 2 grevilleas and 3 jinicuiles are located; while the second at a distance of 1.8 x 2.3, double posture in arrangement three bolillo where 3 huizaches, 3 chalahuites and 3 jinicuiles. The plantation age is approximately 40 years for both varieties.

Sampling of fruits

4 and 7 samples of Mundo Novo and Typica green coffee were obtained, respectively. It is were determined weight, volume and number of cherries per sample. The ripe fruits of the samples were processed by the wet benefit method. It is were pulped separately, fermented, washed, dried and finally threshed for physical and sensory analysis.

Physical and sensory analysis of the samples

The physical analysis and sensory evaluation of the coffee samples was carried out following the protocols of the SCAA (^{SCAA, 2015}), in the tasting laboratory of the Graduate College-Campus Cordoba, located in Cordoba, Veracruz. The physical analysis consisted of: a) determining the moisture content in subsamples of 2.5 g of dust by means of the Mettler Toledo halogen moisture analyzer model HG63; b) classify coffee by its grain size with sieves from number 15 to 19 and base; c) weigh the amount of coffee retained by each sieve for each sample of cherry coffee; and d) calculate the percentage of grains retained per sieve.

The classification by form consisted of separating the triangle, snail, planchuela and elephant grains; while the classification of defects consisted of: a) separating immature grains, bitten, broken, black, sour, etc. in each of the samples; and b) obtain the percentage in relation to the total weight for each of these per sample.

The sensory evaluation consisted of analyzing the coffee samples by four professional tasters, based on the protocol ^{SCAA (2015)}: a) roasting 100 g of each of the cherry coffee samples between 12 and 15 min in a brand roaster Promor; b) determine its degree of roasting when the coffee obtains a light roast; c) weigh 12 g of roasted grain per cup and grind to grade 3 in a Mahlkoning brand mill; d) determine the fragrance of the ground coffee to each of the samples, adding pure boiling water (93 °C) and wait 4 min for the fragrance, aroma, acidity and body attributes to be fully integrated; e) determine the aroma; f) clean the cup, removing with a spoon the foam formed on the surface of the drink; g) determine the acidity, body, uniformity, balance and sweetness of the drink; and h) evaluate sensory attributes with a scale from 0 to 10, where 0 is the lowest and 10 is the highest. The tastings were made per table, each table with 3 samples with five repetitions, where each taster gave the rating of each of the characteristics evaluated. In the end, an average of the tasters’ ratings was made.

Analysis of data

For the analysis of data of the physical evaluation of the varieties under study, a comparison of means was made with the t-Student test (^{Ortiz and Moreno, 2011}). The method consisted of: a) checking the assumptions of homogeneity of variances and normality of errors by means of the Bartlett and Shapiro-Wilk test, respectively; b) apply the Welch test in cases where the assumption of variance homogeneity is not met; and c) perform the t-Student test to compare the means of the variables; that is, evaluate their differences and test the following hypotheses:

Where: μ1 and μ2 is the average percentage of grain retained for the sieve and variety Mundo Novo and Typica, respectively. Ho and H1 are known as null and alternate hypotheses, respectively. This last hypothesis is a bilateral hypothesis, because it can be true if μ1<μ2 or if μ1>μ2 . Hypothesis tests were performed with a level of significance α= 0.05 . Additionally, the Cohen ‘d’ test, also called effect size (TE), was applied in order to measure relatively the superiority of one treatment over another in the population from which the sample is obtained (^{Cerda and Villarroel, 2008}; ^{Ledesma et al., 2008}).

In the data analysis of the sensory evaluation, the principal component analysis (ACP) method was applied, which is a tool that extracts components or factors that better explain the analyzed dimension (^{Johnson, 2000}). Because there was no variability between the varieties, the ACP was applied to identify the components that best explain the quality of the beverage, evaluated through the analysis of sensory quality. This analysis considered the quantitative variables: aroma, flavor, residual flavor, acidity, body, overall appreciation, grain moisture.

Green coffee physical analysis

Table 1 shows the results of the statistics obtained from the physical analysis for both varieties. The average percentage of grain retained in sieve 19 (7.5 mm) was y-1 = 0.057% for Mundo Novo, while for Typica it was y-2 = 12.4%. These percentages indicate that Ho is rejected; that is, the difference between them is statistically significant, for α≤ 0.05, with the t-Student test, and that this difference is also large, according to the Cohen test. Regarding sieve 18 (7.1 mm), the results also indicate that there are statistically significant differences. Hence, the Typica variety retained more grain than Mundo Novo in sieves 18 (28.86% vs. 0.12%) and 19 (12.4 vs. 0.057%). In contrast, in sieves 13 (5 mm), 14 (5.6 mm), 15 (6 mm), 16 (6.3 mm) and 17 (6.7 mm), the results obtained indicate that there are no statistically significant differences in the average percentage retained grain between the two varieties.

Regarding the percentage of grain retained by sieves, Mundo Novo and Typica retained about 50% and 27% of grains classified as medium, respectively, in sieve 16. In this regard, ^{Lopéz-García et al. (2016)} mention that Americans prefer seeds between sieve 15 and 16, while Europeans prefer sieve 17 and 18.

However, ^{Lara (2005)} suggests not to consider altitude, shade and fertilization, since they influence the physical quality of the grain (size, weight and imperfect grains), and organoleptic (aroma, body, acidity, taste and preference). In this sense, ^{Vaast et al. (2005)} mention that with increasing altitude the grain size increases and the percentage of defects decreases. In addition, ^{Marín et al. (2003)} emphasize that grain size is important for the physical quality of coffee, but it is not a criterion of sufficient quality. Therefore, it is necessary to evaluate the appearance and color of almonds. Regarding sieves 13 and 14, there were no grains retained in both varieties.

Regarding the moisture content of the grain in the two varieties, the results showed no statistically significant differences with α= 0.05 . The average humidity percentages were 9, which based on the range (from 10 to 12%) of commercialization humidity of coffee established by the SCAA is low, although for the range (from 9 to 13%) established by Green Coffee Association of New York (GCA), is adequate.

In this regard, ^{Reh et al. (2006)} argue that a percentage of moisture content between 8 and 12.5% is considered adequate, because it avoids problems of microbial growth and mycotoxin formation, which alter the sensory quality in the cup due to the generation of unpleasant flavors. Also, ^{Puerta (2006)} emphasizes that a coffee bean that has not been dried properly develops bad smell and taste, crushes in the thresher and loses a lot of weight during the roasting process, consequently its quality is lower.

Table 2 shows the results obtained from the statistical analysis regarding the shape of the grains. The average percentages of flat grain and snail retained in sieves 19 and 18 between Mundo Novo and Typica varieties are statistically different with a level of significance less than or equal to 0.05, associated with the t-Student test, and this difference is large, according to Cohen’s test. These results coincide with those found by ^{López-García et al. (2016)}, both in the Typica variety and in the Mundo Novo, although, in their investigation they did not obtain significant differences at 95% reliability. In addition, according to ^{Wintgens (2004a)}, flat-type coffee is considered normal for market requirements.

Regarding the grain triangular, the results indicate no statistically significant differences between the means of the two varieties, for a significance level of 0.05, and the difference is negligible as tested by Cohen. These results also coincide with what was found by ^{López-García et al. (2016)} in Arabic coffee varieties. However, there is little or no evidence that this type of grain has a negative effect on the quality of the beverage (^{Wintgens, 2004b}). Under the same level of significance α=0.05 , the average percentages obtained from monster grains and shells do not show statistically significant differences in both varieties. Cohen’s test classified this difference as negligible.

As to the total defects, only secondary defects were presented as a partial black, partial sour philotes and Brocade mild. Their average percentages varieties are not statistically different; that is to say, is negligible according to test Cohen. The foregoing coincides with the results obtained by ^{Marín et al. (2003)}, who found low percentage of almonds with defects in samples collected in mature state.

The results also show that the Typica variety obtained 1.5 partial black defects and 6 complete defects, while Mundo Novo obtained 1 partial black defect and 10 partial sour defects. According to the classification of green coffee SCAA mentioned that part 3 sour or black beans equals 1 full imperfection, Novo Mundo has 3 full imperfections and Typica none. The other secondary defects were not representative, with 1 to 2 defects of mild brocade and floats, respectively.

Sensory quality of coffee

Table 3 shows the results obtained from sensory quality. In general, all the organoleptic indicators obtained average scores above 7, except for the uniformity, sweetness and clean cup indicators that reached an average of 10. In this regard, ^{Puerta (2000)} mentions that the global appreciation is an indicator that allows to accept or reject A sample of coffee for its quality, and is related to all perceived properties with the sense of smell (aromas) and taste (body, bitterness and acidity).

The overall average of the total score of the tasting was 80.7, indicating that coffee is of very good quality and listed as specialty coffee (80-84.99), according to the tasting protocols of the SCAA.

Aromatic descriptors

Table 4 shows the aromatic notes identified by the tasting panel in the coffee samples evaluated, according to the SCAA. Typica and Mundo Novo have the same aromatic subgroups, which coincides with that reported by Hernández (2017), who mentions that the coffee region of Coatepec is cataloged by chocolate notes, accompanied by fruity and caramel notes. These aromatic groups also coincide with what is found by ^{Escamilla (2007)}, with the difference that the notes of nuts, flowers, fruit, phenolic and pyrolytic appear less frequently, and stand out better in coffee trees over 35 years old, average age of Plantations of the farm under study. As for the aromatic notes found and their relationship with the chemical compounds, according to ^{Puerta (2011)}, the fruity and sweet notes are aldehydes, ketones, esters, alcohols and acids, the floral ones are mainly alcohols and the smoky smells are phenols.

Principal component analysis

The ACP allowed grouping the highest percentage of the variability (71.33%) into two components (Table 5) according to the Kaiser Criterion, which assumes that the observable variables have variance 1.

The first component explained 47.20% of the total variability and consists of the variables taste (-0.442), residual taste (-0.417), acidity (-0.412) and global appreciation (-0.485), which are related to the ‘smell and taste’. The second component represented 26.4% of the total variation and was formed by the variables body (0.628) and balance (0.639), which are related to the ‘consistency of the cup’.

In Table 6 and in part a) of Figure 1, the results indicate that the variables are negatively correlated, in component one. The taste is negatively correlated, followed by global appreciation, acidity and residual taste. While in component two, body and balance correlate positively. In subsection b), it can be observed that the varieties of coffee plants evaluated are not determinants in their quality, because they are located interchangeably in the component plane. In this regard, ^{Avelino et al. (2005)} points out that the genetic factor does not influence cup quality; but the quality is more related to the terroir; that is, with the macro climate, which determines the sensory characteristics, including the typicities and the chemical content of the grains.

Figure 1 Main components of coffee sensory attributes.