Analysis Strategy for Assisted Selection of Rice Eating Quality by Using Starch Properties

Abstract


Conclusions
Therefore, the SPQ indicator is a rapid, objective and convenient technique for the rapid assessment of rice eating quality in a large number of crossbreeding progeny, especially early-generation progeny, from which only small samples of rice grains are available.

Background
Rice is the staple food of more than two-thirds of the global population [1]. As countries industrialize and living standards increase, there is more demand for rice with high eating quality [2]. Traditional rice eating-quality measurements mainly rely on sensory evaluations by trained human panels. This is time consuming, labor intensive and prone to subjective errors, because of differences in panelists' geographic origin, age and gender. Sensory evaluations also require large rice samples, and they are di cult to carry out during plant crossbreeding, when only very small samples can be harvested [3]. Given these limitations, extensive studies have been conducted to nd simple physicochemical indicators to replace human sensory evaluations [4,5].
The amylose content of rice has the greatest in uence on its eating quality [6,7]. Rice with high amylose content generally has low viscosity, high hardness, uffy dryness and poor palatability. Reducing the amylose content is a major goal of rice-breeding aiming to improve the palatability of rice [8,9]. However, for those varieties that have similar amylose contents, the eating quality can still vary widely, indicating that other factors also contribute [3]. Rice analysis using Rapid Visco Analyzer (RVA) pro les can characterize physicochemical property changes in rice during starch pasting and shows a signi cant correlation with eating quality [10,11]. The RVA pro le includes determinations of peak viscosity (PV), TV, FV, BD, consistency (CO) and SB. The RVA pro le is closely related to the amylose content and eating quality of rice [12,13], but previous research mostly focused on the comparison of indica or japonica subspecies varieties, and feasible protocols for using the RVA pro le to assist selection for eating quality have not been established [14,15]. Based on the inbred lines of indica-japonica crosses with complex genetic backgrounds, this study aimed to derive a quantitative relationship between amylose content, RVA pro le characteristics and eating quality determined by sensory evaluation, to establish an effective method to assess eating quality by assigning rice starch objective characteristics with numerical values. This would provide technical support for the rapid identi cation of the eating quality of a large number of crossbreeding progeny, especially at the early generation stages.

Results
Performance of rice starch properties and eating quality value in recombinant inbred lines

Relationship between rice amylose content and RVA pro le characteristics
The correlation between AC and TV, BD and SB, reached a very signi cant level of p < 0.01 (Fig. 2) and the correlation with FV reached a signi cant level of p < 0.05. The TV, FV, CO and SB of the RVA pro le characteristics were positively correlated with AC content, whereas PV and BD were negatively correlated with AC content.
Based on the mean AC content of 17.6%, the RILs were classi ed into two groups with high (n = 70) and low (n = 74) AC content. Differences in TV, FV, BD and SB performance were compared in different AC content groups (Fig. 3). The performance of TV, FV and SB in the high AC content group was signi cantly higher than that of the low AC content group, which was opposite to the BD performance; the low AC content group had a higher BD value. Figure 4 shows the correlation between starch properties and EQ values in the RILs. The AC content was negatively correlated with EQ value. Of the RVA pro le characteristics, PV and BD were positively correlated with the EQ value, and the correlation between BD and EQ value reached a very signi cant level of p < 0.05. TV, FV, CO and SB were negatively correlated with EQ value, with a signi cance level of p < 0.01 for TV, FV and SB.

Relationship between starch properties and eating quality in RILs
With EQ values of 60 and 70 as the limiting criteria, RILs were classi ed into high (EQ > 70, n = 48), medium (EQ: 60-70, n = 53) and low (EQ < 60, n = 43) EQ groups. The AC, TV, FV, BD and SB values that were signi cantly related to the EQ value were compared between the three EQ groups (Fig. 5). The high EQ group generally had a low AC, TV, FV and SB, but a high BD. The TV value was the best of the starch characteristics for distinguishing between the medium and low EQ groups.
Analysis of the assisted selection of eating quality of RILs by using starch properties It appeared that the indicators AC, TV, FV, BD and SB were closely related to the EQ value obtained by sensory evaluation. Based on the assessment criteria of EQ, the 25th, 50th and 75th of the distribution of AC, TV, FV, BD and SB values were used as classi cation standards, and the ve indicators were assigned with numeric values i.e., 20, 15, 10 and 5, respectively. The ve indicator scores were added to derive a new indicator-viz., the starch properties quality (SPQ) (Fig. 6A)-which represented the comprehensive characteristics of rice starch that contributed to eating quality. The SPQ and EQ values were positively correlated at the signi cance level of p < 0.001 (Fig. 6B).

Discussion
The TV, FV, BD and SB values obtained from the RVA pro les were signi cantly correlated with AC content. This is consistent with previous research ndings [16,17], but the correlation coe cient in this study was slightly lower than that of indica or japonica subspecies [18]. This is probably attributable to the complex genetic background of indica-japonica hybrid offspring, with more uncertainty regarding in uencing factors [19].
Of the RVA pro le characteristics, the TV, FV, BD and SB values were signi cantly correlated with the EQ value. Using EQ values of 70 and 60 as the limiting criteria to classify RILs into high, medium and low EQ groups revealed that the high EQ group had a pro le of low TV, low FV and low SB, but high BD. Compared with different EQ groups separated by AC content, TV and FV were more signi cantly different between the high and low EQ groups. The TV value can also be used to distinguish between the medium and low EQ groups.
The indicators that were closely related to the EQ value, i.e., AC, TV, FV, BD and SB, were assigned numeric values to derive the SPQ, an integrated indicator that can comprehensively describe the characteristics of starch in terms of EQ. The SPQ was superior to any individual indicator at correlating with the EQ value, at the signi cance level of p < 0.001. Therefore, we conclude that the SPQ value of a starch sample can be used as an accurate assisted-selection indicator of EQ. It requires much smaller rice samples and assessment is much faster. However, before wider application of the SPQ assay in assisted selection, it needs to be veri ed by testing a much wider range of crossbreeding progeny, with diverse genetic backgrounds, to ensure that the correlation between SPQ and EQ can be generalized. Our future work will aim to achieve this veri cation.

Conclusions
In this study, the SPQ indicator was proposed for starch assisted-selection of eating quality holding the advantage of rapid, simple and smaller rice samples. Therefore, this method should be a rapid, objective and convenient technique for the rapid assessment of rice eating quality in a large number of crossbreeding progeny, especially early-generation progeny, from which only small samples of rice grains are available.

Experimental materials and eld trial
The test material was a recombinant inbred line (n = 144) constructed by crossing japonica rice, i.e., Akihikari (AKI) with indica rice, i.e., Qishanzan ( Determination of rice amylose content and RVA pro le characteristics Rice harvested from the eld trial was stored at room temperature for 3 months, after which it was polished and milled into rice our. AC was then determined by using size-exclusion chromatography (SEC) on Ultrahydrogel (Waters, Milford, MA) as described previously [20]. RVA pro le characteristics were determined by 3-D viscosity rapid measurement on a Rapid Visco Analyzer (Newport Scienti c Instruments,Warriewood, NSW, Australia), equipped with TCW (Thermal Cycle for Windows) software for data analysis [21]. For sample preparation, polished rice our (3.00 g, water content 14%), was mixed with distilled water (25.00 ml). During the measurement, the temperature inside the measurement jar was maintained at 50°C for 1 min, then increased to 95°C at 12°C min − 1 , maintained for 2.5 min, then decreased to 50°C at 12°C min − 1 and maintained for 1.4 min. The rotation speed of the agitator was 960 r min − 1 during the initial 10 s and then maintained at 160 r min − 1 . Viscosity values are expressed as rapid visco units (RVUs). The RVA pro le characteristic values mainly include several initial values of PV, TV, FV and their derivatives, including BD (PV-TV), CO (FV-TV) and SB (FV-PV). The meaning of the RVA pro le characteristics was de ned (Fig. 1) based on a previous report [22].
Determination of eating quality The EQ value was determined by sensory evaluation, as described previously [23], but with minor modi cations. Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.      Correlations between individual indicators of SPQ and EQ. A denotes the standard of assigning numeric values for AC, TV, FV, BD and SB in SPQ. B presents the correlation between SPQ and EQ. *** denotes the signi cance level at p < 0.001.

Figure 7
Determination of PV, TV, FV, BD, CO and SB of the RVA pro le characteristics.