The convenience of non-conventional methods for evaluation of the culinary quality of beans

The standard Mattson method to evaluate the cooking time of beans on breeding programs presents disadvantages. We have tested and improved a non-conventional method previously proposed to evaluate the percentage of cooked grains on an automatic pressure cooker. It showed a similar trend as Mattson’s for the cooking behavior of contrasting genotypes, but the first showed a higher correlation with sensory tests. Differently, it is closer to the Brazilian domestic bean preparation. It is a simple and more affordable procedure for application in a huge routine, once it requires a shorter time to run simultaneously a larger number of samples with confident results. Its positive correlation with sensory analysis helps breeders to select promising genotypes from early breeding generations to attend consumer demands more efficiently. We recommend the standard method be used preferably to evaluate the advanced generations to validate the cooking quality of selected materials and to attend cultivar releasing normative.


Introduction
Brazil is one of the main bean producers in world and has one of the highest consumption of beans per capita (CONAB, 2019). Due to seasonality of production in Brazil and in many other countries, storage of beans is usual and necessary to maintain bean supply throughout the year, avoiding scarcity between harvests (Maia, Pinto, Marques, Lyram, & Roitman, 2013). However, inappropriate storage causes undesirable changes on final product, leading consumers to reject it (Alvares, Pereira, Melo, Miklas, & Melo, 2020;Carbonell, Chiorato, Gonçalves, Perina, & Carvalho, 2010;Scariot, Tiburski, Reichert Júnior, Radünz, & Meneguzzo, 2017). During conventional storage, when conditions are not controlled, carioca beans tend to develop undesirable characteristics in terms of quality. Among these characteristics the darkening of tegument is the parameter that consumer associates with a longer cooking time and considers that product is old (Alvares et al., 2020;Bento et al., 2020;de Farias, Devilla, Silva, Bento, & Bassinello, 2020;Rupollo et al., 2011;Vanier et al., 2019).
To evaluate the beans cooking time, the most adopted method in Brazil is the one recommended by the Brazilian Ministry of Agriculture, Livestock and Food Supply (MAPA), which uses the Mattson cooker apparatus. However, this technique presents many disadvantages and limitations such as: it is time-consuming; it causes fatigue to the operator (based on prolonged visual evaluation) and can lead to errors or limited samples to be daily evaluated; it has low work capacity per day, which is particularly a big problem for the great demand of breeding programs that need quick answers for programming new crosses. The interpretation of the result of the standard method is an estimate of the grain resistance to cooking and the cooking time generated is not directly compatible with the domestic method of preparing the grain and has a low correlation with sensory analysis, regarding texture or cooking quality acceptability (Carvalho, Ramalho, Vieira Júnior, & Abreu, 2017). Therefore, it is clear the importance of assessing new, more agile, and practical alternatives to evaluate the culinary quality of beans, to meet the demand of the breeding programs.
In this sense Carvalho et al. (2017) has proposed a procedure for cooking beans in an electric pressure cooker in a predefined time and then evaluate the percentage of cooked grains using Mattson apparatus only as a kind of penetrometer tool with a needle-type probe to easily and rapidly estimate the totally cooked grains. Even though it still uses the Mattson apparatus at the end of analysis to verify the cooked grains, this check is practically immediate and so it increases the analysis' yield of each equipment. The main goal was to get reasonable cooking quality data by reducing the analysis lead-time and increasing the number of samples in routines Development, v. 9, n. 11, e44491110103, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i11.10103 5 of bean analysis demanded by genetic breeding programs. Besides, the method, independently of pressure cooker variation between commercial brands, is the closest to domestic preparation procedure and can simulate what consumers could really find by cooking beans. The alternative method also allows to cook different samples at the same time under the same cooking conditions and to check which ones could be considered properly cooked based on its grain softness instead of using the cooking time as a quality parameter. The most important interest, especially when evaluating early breeding generations, is to segregate low cooking quality beans from the high-quality ones in a quick way. However, the authors had not defined the percentage of cooked beans considered acceptable or ideal (cutoff) when selecting promising lineages or cultivars, once they had not run sensory tests or even instrumental texture analysis as reference, neither an evaluation of method efficacy aiming a potential replacement of standard method by research centers.
Due to such aspects, the objective of the present work was to test the proposed nonconventional method in order to confirm its viability in routine analysis and its data compatibility in comparison to standard and other alternative methods; to define its cutoff value for percentage of cooked beans based on sensory evaluation to help breeders with the genotype selection as well as to check its capacity of differentiating the cooking quality of freshly harvested and stored carioca beans with known different darkening and hardening profile.

Material
Bean cultivars from commercial carioca group, Perola, BRS Estilo and BRSMG Madreperola (Figure 1), were produced at experimental field of Capivara farm of Embrapa Rice and Beans, in the municipality of Santo Antônio de Goiás, State of Goiás, Brazil, using a randomized block design, with 8 rows of 4 meters each, and three replicates. All genotypes received the same treatment during handling and post-harvesting. Samples were analyzed in three periods (freshly harvested, at 3 months and at 6 months) during the six months of storage.

Sample preparation
After harvest, beans were treated for pest control, selected only whole and healthy beans, dried by natural drying process under room conditions and then cleaned. Samples were quartered and stored in a cool, dry and non-air-conditioned room in sealed polyethylene plastic bags with 1 kg of capacity. Storage was carried out in a room with natural light from the rising sun, where temperature and relative humidity were recorded using a digital thermohygrometer (Table 1).

Parameters for Bean Color
The color of raw (RW) and cooked beans (CO) was determined in a Hunterlab colorimeter (Colorquest), which considers the Cielab system (L* a* b*), responsible for luminosity and intensity of red and yellow color, respectively. Results obtained were presented in terms of luminosity (L*), chromaticity a* and b*, and total color difference (∆E*) as described by Bento et al. (2020).

Standard method for determination of bean resistance to cooking by Mattson cooker (RMC)
Classification of bean resistance to cooking was based on a scale defined by Proctor and Watts (1987) after cooking beans in Mattson cooker apparatus. The time (t13) recorded until the drop of the 13 rd rod is converted into a rank (RMC) of resistance to cook as described by Bento et al. (2020). Research, Society and Development, v. 9, n. 11, e44491110103, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i11.10103

Percentage of Cooked Beans in Electric Pressure Cooker (PCEPC)
This parameter was determined according to the method proposed by Carvalho et al. (2017) with modifications. Each cultivar replicate of 50 grains were sealed in net sachets and soaked in distilled water at room temperature for 60 minutes. Then, they were cooked with new amount of distilled water (a 1:5 ratio) in an electric pressure cooker (EPC -Mondial brand) for 20 minutes. The cooked grains were drained, cooled down up to room temperature and then put at the base of Mattson apparatus so that the rods were set over them simultaneously and the percentage of fully and immediately penetrated beans was counted ( Figure 3).

Instrumental hardnesstexture
Beans cooked in electric pressure cooker -EPC -and electric plate -EP -were evaluated for instrumental hardness, with adaptations according to Revilla and Vivar-Quintana (2008). For cooking in EPC, beans were soaked for 60 minutes and then cooked for 20 minutes in water in a 1:5 ratio. The beans cooked in EP were soaked for 18 hours and cooked in water in a 1:5 ratio. Heating temperature varied from 290 to 300 °C and cooking time was determined according to the time found at Mattson apparatus. After bean cooking, only whole beans were selected for hardness test, each bean was placed on a plate of TA-XT2i texturometer (Stable Micro Systems, Surrey, Tax Plus), using a compression force of 80% and a cylindrical probe at a constant velocity of 1.0 mm s -1. Hardness was evaluated using a 2 mm diameter probe.

Percentage of damaged beans after cooking (DB)
After cooking in EPC and HP, the total mass of beans was manually separated into two portions, damaged and undamaged. The seed coat detachment, the presence of cracks in seed coat and presence of split beans were considered for classification. Then, beans had their mass determined by dividing the number of beans damaged by the total analyzed. Results were expressed as a percentage.

Sensory Analysis
Sensory analysis was carried out using a nine-point hedonic scale acceptance test, according to the method proposed by Stone, Bleibaum, and Thormas (2012), in freshly harvested and stored raw beans to study the acceptance of overall appearance, color and intention of purchase of consumers along the bean storage. For sensorial analysis of cooked samples (in electric pressure cooker), the ideal scale model was used to classify the degree of hardness, color and appearance of freshly harvested and stored beans. Analyzes were performed by at least 60 untrained testers, in individual booths under white illumination. The project was approved by Plataforma Brasil Ethics Committee with number CAAE 44815115.9.0000.5083.

Statistical Evaluation
Quantitative results were obtained in three replicates, analyzed for their normality, then submitted to the Kruskal-Wallis test, and later compared by Wilcoxon test at 5% significance Research, Society and Development, v. 9, n. 11, e44491110103, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i11.10103 level. The effect of storage time was studied using regression analysis. Statistical tests were performed using RStudio free software.

Color parameters
The score rankings of visual commercial color scale (VCS), the instrumental values for luminosity (L*) and chromaticity a* and b*, and the score frequency distribution of sensory analysis from raw carioca bean cultivars are presented in Table 2. Beans from cultivar BRSMG Madreperola were the best classified according to the VCS, with notes between 9.33 (after 6 months of storage) and 10.0 (freshly harvested), and presented the highest scores by sensory panel, which classified beans as light-colored or very light-colored. The darkening due to storage was noticed in beans stored for 6 months by only 16.12% of tasters with a 5 score (slightly light-colored). However, there was no variation in the score attributed by Bolsinha color scale between the third and the sixth month of storage. Regarding the instrumental color data from BRSMG Madreperola, the luminosity (L*) presented values between 54.15 (freshly harvested) and 52.82 (6 months of storage), a significant variation (p <0.05). On the other hand, Perola and BRS Estilo had a score by visual scale below 8, except freshly harvested beans from BRS Estilo. The sensory panel scores ranged between 4 (neither light nor dark) and 1 (very dark), where beans stored for 6 months received the lowest scores (darker) by 54.83% and 68.06% (Perola and BRS Estilo, respectively). Darkening during storage was also detected by instrumental color measurement, where L* values decreased, and chroma a*, increased.  Figure 1); L*: brightness; a* and b*: chromaticity. Source: Authors. Table 2 shows that L* values above 52.82 ± 2.09, and chroma a* and b* lower than 7.54 ± 0.23 and 19.16 ± 0.23, respectively, coincide with visual scale scores above 9, and higher scores frequency (> 67%) from sensory panel corresponding to light-colored beans. However, when evaluating color of cooked beans, no significant difference is observed among cultivars.
Storage time did not influence coloring of cooked beans from cultivars Perola and BRS Estilo.
On the other hand, BRSMG Madreperola presented an increase of yellow color (chroma b*), increasing the color difference (ΔE3,6) during storage period (Table 3). Depending on the cultivar, the color of seed coat is a characteristic that influences the final consumer's choice, while browning is often associated with longer storage and cooking time (Coelho, Prudencio, Nóbrega, & Leite, 2009).
Research, Society and Development, v. 9, n. 11, e44491110103, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i11.10103 Table 3. Luminosity, a* and b* chromaticity and total color difference (ΔE) of colorimetric evaluation, and sensorial evaluation of tegument color uniformity of cooked beans from different cultivars, freshly harvested (0) and after 3 and 6 months of storage. According to the Bolsinha (2014), the market negotiates the price of carioca beans taking into account the color on a 5-10 classification scale, where beans scored as 7 are darker than those scored as 10. This classification system (not officially adopted by MAPA) is a method based on visual inspection and is the most widely used in bean marketing, mainly due to its quite simple procedure. On the other hand, the accuracy of test depends on evaluator's expertise and so, presents subjective results besides the influence of inherent commercial Research, Society and Development, v. 9, n. 11, e44491110103, 2020 (CC BY 4.0) | ISSN 2525-3409 | DOI: http://dx.doi.org/10.33448/rsd-v9i11.10103 13 interests. One alternative is the use of a colorimeter, inclusive its portable version, which provides more precise results for sample L*values (how much light or dark the sample is).
Based on our data, it can be inferred that instrumental color evaluation is the most indicated to provide a better distinction among different genotype colors, additionally to the fact that is not a subjective analysis. As an example, the Bolsinha' brokers from São Paulo and sensory panel scoring could not detect any color difference for BRSMG Madreperola from three to six-month-storage. However, instrumental analysis did show a difference (ΔE6 = 2.39), which is considered significant (Ribeiro, Storck, & Poersch, 2008).

Resistance to cooking, percentage of cooked bean, bean hardness and percentage of damaged beans
Based on the evaluation of bean resistance to cooking in Mattson apparatus (RCM), it was verified that all cultivars recently harvested showed normal resistance to cooking (Table   4), with a slight increase of cooking time during storage, except BRSMG Madreperola that had a significant increase (50.07%) after the third month, being classified as very resistant to cooking (42.26 minutes) at the end of storage.
According to the method adapted and proposed by Carvalho et al. (2017), the percentage of cooked beans in electric pressure cooker (PCEPC) for freshly harvested cultivars was high (81% to 97%) and there was no statistical difference between Perola and BRSMG Madreperola (p > 0.05) (Table 4). At third month, all cultivars differed statistically (p <0.05) between each other, but all had significant reduction in the percentage values. Finally, at sixth month, all cultivars presented an additional reduction in the percentage of cooked beans with no statistical differences from each other, and this reduction was more pronounced (34.6%) for cultivar BRSMG Madreperola (96.67% to 63.20%).
Instrumental hardness of freshly harvested beans when cooked in HP showed statistical difference (p < 0.05) between Perola and BRS Estilo, but during storage the cultivars did not present a statistical difference (Table 4). For cooked beans in electric pressure cooker, it can be observed that freshly harvested beans do not present statistical difference (p < 0.05) and during the storage time, samples of BRSMG Madreperola became harder than others (Supplementary Table 2). There was no presence of damaged beans when cooking in HP. Considering the percentage of damaged beans cooked in EPC, freshly harvested beans of Perola presented the highest value (12.51% ± 1.25) with no significant difference to BRS Estilo (11.12% ± 0.54), followed by BRSMG Madreperola (9.04% ± 0.31). After six months, BRS Estilo remained as the cultivar with the highest percentage of damaged beans, differing statistically from the others (Table 3).
From Table 4, it is possible to check the hardness increase tendency of all cultivars during storage, however in different intensities or ratios, depending on the method and the genotype. By comparing the capacity of hardness differentiation of both methods (Mattson and modified non-conventional method) between Perola and BRS Estilo up to 6 month-storage, we can notice that the modified method seemed to be more sensitive. According to Mattson's, Perola had no change of its cooking resistance classification along storage period, keeping as normal, and BRS Estilo changed only at the sixth month, becoming moderately resistant; while based on the modified method, both cultivars had significant reduction of the percentage of cooked grains since the first months of storage (24% reduction for Perola and 18% reduction for BRS Estilo at the third month-storage). It is important to highlight, in addition to the greater sensitivity of the modified method, its negative correlation to the instrumental texture ( Figure   4), which in fact indicates the same profile of hardening increase by both, as well as they are in accordance with the sensory analyses, that also verified an increase of the perception of hardness in the stored grains. Regarding cultivar BRSMG Madreperola, when comparing resistance to cooking in Mattson apparatus with the percentage of cooked beans, it can be observed that beans recently harvested and stored for up to three months showed about 90 ± 6% of cooked beans, which is approximately 29 ± 1 minutes in Mattson scale, while the beans stored for 6 months demanded an average cooking time of 42 min, corresponding to approximately 60% of cooked beans in electric pressure cooker. In other words, both methods indicated that the beans grains harden, which is in accordance with sensory panel and the results of texture analysis since they also detected the hardness increase (Table 4).
In order to stipulate a reasonable value for genotype selection based on its cooking quality measured by the modified method, we have considered the percentage of cooked beans which had more than 50% of acceptability by tasters (who classified cooked beans as soft/ideal or with low hardness). So, by adopting 70% as the cutoff value for the modified method, we guarantee that samples with low, normal, and moderate resistance to cooking will be selected by breeders, as it might be by using Mattson cooker. Besides, the modified method showed some other advantages, such as the shorter time required for each sample test (average time per analysis is about 1 hour and 30 minutes, considering the soaking time of 60 min, the cooking time of 20 min, the 6 min for letting beans to cool down, and finally, the time to put the grains in Mattson apparatus for the immediate counting of rod penetrated grains), the higher capacity for multiple analysis in one single test (up to 35 samples can be cooked simultaneously and be rapidly further analyzed) and its cooking conditions, simulating domestic bean preparation which makes it closer to real conditions of consumers. On the other hand, we consider that, even having reduced the soaking time before cooking (in comparison with Mattson's) as well as the cooking time set in pressure cooker, these parameters could be further optimized to improve bean quality evaluation, and this adjustment is dependent on the genetic breeding goals and the market demands, as well as we had adjusted the cooking time in the pressure cooker. It is known that modern consumers tend to abort the soaking step before cooking, as new cultivars available in supermarkets are softer and cook faster (Rodríguez-González & Fernández-Rojas, 2015), as a result of genetic breeding for this quality trait.
The negative correlation presented between the method adapted by Carvalho et al. (2017) and the traditional one, indicates the accuracy of data, as beans that showed greater resistance to Mattson cooking method has a low percentage of cooked and damaged beans.
Another point that should be taken into account when using Mattson method is the intrinsic data variation, requiring the use of a large number of repetitions of the same experimental trial   beans also observed that cooking time (bean resistance to cooking) gets longer as the storage period increases (Coelho et al., 2009;Silochi et al., 2016). For carioca beans, only two months of storage are sufficient to observe a hardness increasing, even under controlled conditions (5 °C) (Coelho, Prudencio, Christ, Sampaio, & Schoeninger, 2013).
In summary, the non-conventional method to determine the cooking quality of beans using an electric pressure cooker (soaking the beans for 60 minutes before cooking for 20 minutes under pressure) showed promising results and may be recommended for plant breeders in order to select competitive lineages for the Brazilian market, especially those from early breeding generations. It does not matter at this point if there can be samples among them that would cook in a shorter time than that set in the method, because certainly, both, the best and the good ones, will be selected by this tool to continue on the genetic crossings. Finally, the cutoff value might be 70% of cooked beans, when more than 50% of the tasters should classify the beans as having ideal or low hardness.

Conclusion
The proposed method for evaluating bean cooking quality has shown to be promising, as it reproduces usual cooking conditions adopted by Brazilian consumers, it is almost 12 times faster when compared to the usual method and does not depend on the analyst's full attention during the whole test time. Moreover, the modified method proved to be more sensitive to identify aged or hardened grains than the standard one. Similar situation was observed for color analysis, in which instrumental method presented greater repeatability and sensitivity, with a shorter analysis time and no need of a trained sensory panel. A cutoff value established for percentage of cooked beans obtained from electric pressure cooker was 70%, when more than 50% of the tasters classified the beans as soft or ideal grains. In conclusion, the nonconventional method brings a great contribution to the bean breeding programs as an affordable routine procedure to assist the fast selection of good cooking quality genotypes, especially from segregating populations, with reliable data, which in turn, better match domestic preparation and directly impact the market acceptability much more efficiently than the old standard method. Finally, the next steps in this study will be to expand the testing of unconventional methods in a greater number of carioca bean cultivars and to include colored grain cultivars.