Carioca bean genotypes for tolerance to grain darkening by natural and accelerated methods

The slow darkening of grains is sought by bean breeders because the consumers consider that darker grains demand more time for cooking. The analysis currently used takes around 90 days to differentiate grain color among genotypes. The objective was to evaluate the color as a function of the value of L* (lightness) of carioca beans, by natural and accelerated methods to verify equivalence between methods, validation of the methodology and identification of genotypes tolerant to the darkening. The grain darkening was compared and evaluated by natural darkening method under shelf conditions, in days storage, and accelerated darkening method under ultraviolet light, in hours. The natural darkening time of 90 days was statistically equal to 24 hours of accelerated darkening, and the difference among the genotypes could be obtained in a shorter time, indicating a correspondence in the methods. The accelerated darkening method can be used to shorten the analysis time in the routine of breeding programs.


INTRODUCTION
The darkening of the tegument of bean grains (Phaseolus vulgaris L.) is an undesirable characteristic for the consumers because it is associated to longer cooking time (Silva et al. 2008) and old bean. This association is not correct because the darker grains do not always lead to longer cooking times or are the oldest ones. The darkening of the grains is influenced by the environment (humidity, temperature and harvesting time), seasons and genotype (Araújo;Ramalho;Abreu, 2012;Silva et al. 2008). In the harvest, rain plays an important role, since it affects the quality of the grain, which according to Carbonell, Carvalho and Pereira (2003), causes membrane permeability. Excess moisture accelerates fermentation of the grain, making it dark, but not necessarily old.
Currently, the standard methodology for evaluation of the grain used is the natural darkening, which consists of packing the beans in zip lock bags and storing them under shelf conditions at environmental temperature and humidity, with coloring scouting every thirty days. The color tone was given as a function of the value of L* (lightness) (Ribeiro et al., 2014), using a colorimeter, but there are also studies that use scale notes (Alvares et al., 2016) by comparison with other control cultivars. The value of L* has often been used for the simplicity of obtaining and interpreting the results compared to a* (red/green) and b* (yellow/ blue) coordinates. This natural darkening methodology enables the identification of cultivars with slow darkening and needs approximately 90 days to show significant differences among the genotypes.
Strategies have been adopted for more rapid and equivalent evaluation of darkening of grains. Siqueira et al. (2016) worked with accelerated aging, in which the grains were aged in the hot air furnace dark at 40±5 °C and 75% relative humidity to accelerate the darkening process. Artificial methods with ultraviolet light illumination have emerged as an alternative to decrease the time needed to test bean darkening and to establish a methodology for the routine of breeding programs. Thus, the accelerated darkening method under ultraviolet and fluorescent light exposure developed by Junk-Knievel, Vandenberg and Bett (2007) was adapted in relation to the equipment installations (chamber) and sampling time, measuring the value of L* every 24 hours, at 0 hour, 24 hours, 48, hours and 96 hours, as it provides faster results in comparison to natural darkening.
Thus, this study aimed to evaluate grain darkening of 19 bean genotypes, including lines and cultivars of carioca tegument, by natural and accelerated (with ultraviolet light) methods using ANOVA, MANOVA, canonical discriminant analysis, contrasts, Tukey and correlation to verify equivalence between methods, validation of the methodology and identification of genotypes tolerant to the accelerated grains darkening.
The trials were carried out in three seasons: "dry" (Campinas and Tatuí), "winter" (Votuporanga and Ribeirão Preto) and the "rainy" (Mococa and Campinas) in the state of São Paulo, Brazil in 2016. The experimental in the six environments was designed for randomized blocks with three replications. Plots consisted of four 4 m length rows, spaced at 0.50 m, with 10 to 12 viable plants per meter, the useful area of the plot being the two center rows.

Traits analyzed
The genotypes were evaluated based on the value of L* (lightness) of grain tegument by the natural darkening method under shelf conditions (Alvares et al., 2016) and accelerated darkening method under ultraviolet and fluorescent light chamber conditions (Junk-Knievel, Vandenberg;Bett 2007). The grains harvested were submitted to the same conditions of temperature, light and humidity before the analyzes. The values of L* by the CIELAB system, which represent the lightness scale from 0 (black) to 100 (white). Data was recorded using a colorimeter (model CR-410, Konica Minolta, Osaka, Japan) and expressed by the mean of five measurement repetitions for each sample. The standard illuminant D65 was used (corresponding to daylight, including ultraviolet light) and 2 nd standard observer. This value was presented in the form of a unit of measurement (u.m.) of the parameter.
(a) Natural Darkening Method of the Grain: The grains were packed in zip lock bags (8.5 x 12 cm, 80 grams each) and stored under shelf conditions (Alvares et al., 2016) with environmental temperature and humidity, in a room at Santa Elisa Farm -IAC, Campinas (mean annual temperature: 21.4 ºC, mean annual relative humidity: 71%) ( Figure 1A). The gradual change in the color of the grains of the genotypes was determined at 0, 30, 60 and 90 days. During storage, the grains were moved within the bag so that all were exposed to the same lighting conditions; and arranged entirely at random on the shelves at a least once a week.
(b) Accelerated Darkening Method of the Grain: The methodology was adapted from Junk-Knievel, Vandenberg and Bett (2007). Two ultraviolet lamps (wavelength centered at λ = 253.7 nm, model TUV 36W/ G36 T8, Philips, Holland) and two fluorescent lamps (λ = 480 nm, model TL 40W/75RS, extra daylight, Philips, U.S.A.) were installed alternately in the Biotronette Mark III environmental chamber, n. 846 (Lab-Line Instruments, Inc., Illinois) ( Figure 1B). The grains were placed in high transparency polystyrene Petri dishes, 90 mm in diameter by 15 mm in height (model K30-9015O, Olen, Kasvi, Brazil), eighteen centimeters below the lamps for 96 hours. The irradiance of ultraviolet energy Ciência e Agrotecnologia, 43:e012519, 2019 (0.95±0.1 mW cm -2 ) in the surface of the samples was measured using a standard photodiode power sensor, ultraviolet extended (Model S120VC, 200-1100 nm, Thorlabs, Inc., USA). Adding ultraviolet and white light (4.0 ± 0.2 mW cm -2 ), the samples were exposed to a total irradiance of 5.0 mW cm -2 . After 24 hours (86400 seconds) of illumination, the total light fluence calculated was approximately 432 J cm -2 and in 96 hours it was 1,728 J cm -2 . Ultraviolet represents approximately 20% of total light fluence, resulting in 82 J cm -2 in 24 hours and 328 J cm -2 after 96 hours. The value of L* was measured every 24 hours to evaluate the pattern of grain darkening. The volume of grains corresponded to a non-overlapping layer in a Petri dish (± 25 grams, 90 grains), mixed at every color tone reading and placed back in the chamber completely randomly. The ventilating fan remained on throughout the experiment and the temperature corresponded to ± 37 °C.

Statistical analysis
The values of lightness (L*) were submitted to univariate analysis of variance by environment and a joint analysis of variance was performed using the procedure for general linear models (Proc GLM), after verifying the residual mean square magnitudes. All effects, except the error, were considered as fixed. The means of the genotypes were compared by the Tukey test, at 5% level of significance, Pearson's phenotypic correlation coefficients were estimated among the traits. It was also performed multivariate analysis of variance (MANOVA) and canonical discriminant analysis, in order to compare methods, genotypes and environments, and to show the greatest possible separation among them jointly considering all the traits, namely: the lightness at the darkening times 0, 30, 60 and 90 days; 24, 48, 72 and 96 hours. The analyzes were performed using the SAS statistical software (Version Studio, SAS Institute, Inc. Cary, NC).

RESULTS AND DISCUSSION
In the present work the grain darkening protocol was adapted by ultraviolet light established by Junk-Knievel, Vandenberg and Bett (2007) for carioca bean. Regarding equipment installations and sampling time, the present study aimed to show the advantages in the substitution of the current methodology of evaluation of natural grain darkening by accelerated darkening.
The method introduced for accelerated grain darkening was used by assembling a system with minor modifications of ultraviolet and white fluorescent lamps and conditions, such as distance from the samples to the 10 cm to 18 cm lamps. The irradiance measurement showed a significant difference compared to the 4 mW cm -2 of ultraviolet irradiance previously published by Junk-Knievel, Vandenberg and Bett (2007). The total light fluence represents the total amount of photons of ultraviolet and visible light that passed through the area where the beans were placed. In order to calculate the total light fluence (J cm -2 ), it first measured the power per area unit (mW cm -2 ) using a power meter (previously described) and multiplied the value obtained by the time of illumination (seconds).  Table 1 shows the joint analysis for the value of L* (lightness) in bean genotypes of the carioca type, with the coefficient of variation (CV) and the mean of the traits. The joint analysis was performed after verifying the homogeneity of the variances. In this way, no environment influence was excluded and the data were analyzed together.
The genotypes and environments showed significant effect, indicating that the genotypes present significant differences among them for the value of L* in the darkening times evaluated, as well as significant difference among the evaluated environments. The results of the coefficients of variation (CV) obtained were considered low (<10%) according to Pimentel-Gomes (2009) and ranged from 1.78% (72 hours) to 2.47% (90 days). The CVs indicate good experimental accuracy. The proximity of the mean value of L* for 90 days (48.67) and 24 hours (48.46) is highlighted.
The mean values of L* obtained at 90 days did not differ statistically from the values obtained with 24 hours in the accelerated method by Tukey test (5%), which shows a similar performance of darkening in these two methods ( Figure 2). The Pearson correlation estimate based on the residues among traits 90 days and 24 hours was of mean magnitude (r = 0.55), positive and significant at 1%. At 90 days it was possible to verify genetic variability among the genotypes, as they lost color according to the values of L*, which makes the association important in terms of hours for this evaluation. At 120 days the grains were dark, making it difficult to identify genetic differences among the genotypes with the same darkening pattern, as observed by Siqueira et al. (2016). The profile analysis over time was performed between times 1 (0 day / hour), 2 (30 days / 24 hours), 3 (60 days / 48 hours) and 4 (90 days / 72 hours).
According to the multivariate analysis, there is a significant difference at 1% between the profiles in relation to the two levels studied (natural and accelerated darkening). This difference started to be more evident after time 2.
The effect of genotype x environment interaction was significant (Table 1), agreeing with Ribeiro, Jost and Cargnelutti Filho (2004) that evaluated the value of L* in carioca beans. Thus, the performance of each genotype for darkening does not follow the same pattern, depending on the environment in which it was evaluated. This is due to the difference among the environments, with the presence of rain or not at the time of harvest, as well as high temperatures.  However, Junk-Knievel, Vandenberg and Bett (2007) reported that they obtained results of the effects of genotype x environment interaction not significant for the beans with pinto type. In the present work a greater number of genotypes from different research institutions were used, considering also that the trials were carried out in different environments in regions of tropical climate, contributing to a higher occurrence of genotype interactions per environment. This explains why in Brazil studies of tolerance to grain darkening are important in breeding programs.
The main effects of genotypes and environments were explored considering that the coefficients of variation obtained were low, as well as the presence of residual squares of low magnitude, which leads to a smaller contribution of the error to the differences among the genotypes. The classification of genotypes was similar and the interactions were mainly due to differences in magnitude, where darkening was more pronounced depending on the environment in which the genotypes were cultivated. Thus the contribution to the interaction genotypes by environments was mainly due to genetic rather than environmental effects.
The literature presents the genetic control for grain darkening as monogenic or oligogenic, but without consensus. Junk-Knievel, Vandenberg and Bett (2008) and Silva et al. (2008) suggested that there is a single recessive gene that controls the phenotype of slow darkening. Elsadr et al. (2011) and Silva et al. (2014) presented as oligogenic, with the presence of epistasis, in which the expression of a gene depends on the action of a gene other than one of its alleles. As the genotype x environment interaction was significant in the present study, it is considered that the trait is oligogenic or even polygenic, although this trait was not addressed in any of the analyzes performed.
The Tukey test at 5% of significance (Table 2) was performed based on the joint analysis of the data, to compare the means of the 19 common bean genotypes for value of L* conducted in six environments. On average, the lighter genotypes were Gen 45-2F-293P, Gen 4-1F-19P and Gen 107-14A-336 and the darker BRS Pérola and CHC 01-175-1. The genotype CHC 01-175-1 has in its genealogy the BRS Pérola as one of the parents, which explains the strong influence in the darkening. According to Siqueira et al. (2016), darkening in lighter genotypes seems to be mainly due to the polyphenol oxidase activity, whereas in the dark there is a combination of enzymatic and non-enzymatic oxidation.    Junk-Knievel, Vandenberg and Bett (2007) suggested that grain darkening is associated with the presence of proanthocyanidins (condensed tannins) present in the tegument. According to Duwadi et al. (2018), most flavonoids, including catechin monomer proanthocyanidins, accumulate at higher levels in regular darkening than slow darkening genotypes. Proanthocyanidins are oligomeric flavonoids composed primarily of catechin and epicatechin units (Duwadi et al., 2018). Table 3 shows the analysis of the environments, where it was observed that, in general, the environment that presented lighter grains was Campinas, both in the "dry" and "rainy" season. Minimum significant differences ranged from 0.43 (72 and 96 hours) to 0.65 (60 and 90 days). The programs have converged to the improvement to tolerance to darkening of grains and the evaluated genotypes showed low variation, as observed in Tables 2  and 3, where the observed differences are mainly due to genotypes rather than environments.
According to García-Peña and Dias (2009), generally, when there is more than one trait measured per plot in the experimental designs, the multivariate analysis of variance (MANOVA) should be performed. Thus, the analysis was performed and the tests Wilks' Lambda, Pillai's trace, Hotelling-Lawley trace and Roy's greatest root for each of the effects (genotypes, environments and interaction) were calculated. From the results of the four multivariate tests (F = 2.74; F = 2.34; F = 3.27 and F = 17.13 respectively, significance at 1%) the null hypothesis is that there is no difference among the genotypes, H 0Gen =Gen 1 =Gen 2 =…=Gen g , wherein g=19 was rejected. This indicates that there is a difference among the genotypes considering the eight traits responses of value of L* at 0, 30, 60 and 90 days (natural darkening) and 24, 48, 72 and 96 hours (accelerated darkening) simultaneously.
The null hypothesis is that there is no difference among the environments, H 0Environment =Environment 1 = Environment 2 = ... =Environment e , wherein e=6. The null hypothesis was reject based on the tests Wilks' Lambda, Pillai's trace, Hotelling-Lawley trace and Roy's greatest root (F = 33.44; F = 23.14; F = 48.52 and F = 170.06 significant at 1%), indicating that there is a significant difference among the environments considering the eight L* responses traits evaluated jointly.
The null hypothesis of no interaction effect was tested H 0GenEnvironment =Ge 11 =Ge 12 = ... =Ge Ge and rejected by the four tests (F = 1.24; F = 1.20; F = 1.28 and F = 2.99 significant at 1%). It indicates, therefore, that there is a significant interaction effect of the genotypes and the environments on the response traits.  From the results of MANOVA that presented significant values for the genotype and environment factors, the contrasts analysis was performed to identify which genotypes and environments present significant differences in the means of the vectors. The effects of genotypes and environments were highly significant for the same contrasts and are presented below for value of L* considering all the darkening times in both natural and accelerated methods.
Considering the results of the comparison contrasts of the mean vectors from MANOVA and canonical discriminant analysis, Gen 45-2F-293P genotype was distinguished with contrasts with dark genotypes, especially with BRS Pérola. This shows the potential of the genotype to maintain the lightness of the tegument over time, an important feature of tolerance to darkening.
The IAC lines presented contrast with the BRS Pérola, which is an old cultivar in the market, which shows that the breeding program is being conducted satisfactorily for the trait. For bean genotypes with carioca tegument, genotypes with a value of L* higher than 55 are desirable, which is associated with lighter grains (Ribeiro et al. 2014;Siqueira et al. 2016) at the time of harvest. On the other hand, lighter grains tend to be more susceptible to the pathogen Fusarium oxysporum (Chiorato et al., 2015) and to other diseases and pests. This relationship has been observed in field conditions, however, specific studies are necessary to evaluate the selection threshold between light and dark grains.  In the Genetic Bean Breeding Program of the Instituto Agronômico (IAC), the ideal value of L* used has been 53 at the time of harvest as a fast selection criterion, combining market favoring, disease tolerance and broth quality. Below this value the market refutes the product, but there is greater tolerance to diseases and better quality of broth (thick broth). Above this value, the market accepts the product, where, however, a greater number of genotypes susceptible to soil diseases are identified, as well as a lower quality of broth (clear broth).
The polyphenolic compounds are related to the plant defense system and high levels of these compounds are responsible for plants more resistant to pest attack (Islam et al., 2003), however, these genotypes generally present more rapid darkening process of the grain. According to Erfatpour, Navabi and Pauls (2018), polyphenolic compounds have been associated with the color and pattern of grain coverage in common beans. Slow darkening is significantly associated with reduced levels of kaempferol and polyphenol oxidase activity, which is responsible for the oxidation of polyphenols (Beninger et al., 2005;Duwadi et al., 2018).
With the growing appeal for the lower use of pesticides, darker grains that tend to be more tolerant to pests and diseases might respond to market niches focused primarily on organic management. Chen et al. (2015) presents results that confirm the relation of the polyphenolic compounds with darkening and makes a revision on the importance of these and compounds contained in the bean.
The darkening genotypes that slowly presented similar results considering the two methods (Table 2), conferring reliability on the choice of accelerated darkening protocol in relation to natural darkening. Erfatpour, Navabi and Pauls (2018) modified the time of exposure of pinto beans to ultraviolet light from 72 hours to 24 hours but did not present the reason for the time reduction. The time correspondence of 24 hours and 90 days, as well as the non-existence of significant differences by the Tukey test (5%) could allow the breeding programs to implement the accelerated darkening method with ultraviolet light as a standard in the routine.

CONCLUSIONS
The evaluation of the accelerated grain darkening method based on the values of L* (light grains) under ultraviolet light chamber conditions can be used to identify bean genotypes tolerant to grain darkening in 24 hours.