A study on grading of quantitative characteristics of litchi

Abstract：【Objective】The DUS test guideline is an important technical standard for DUS testing, in which the characteristics are divided into qualitative characteristics, quantitative characteristics and pseudo-qualitative characteristics according to their expression states. However, the nine quantitative characteristics in litchi (Litchi chinensis Sonn.) DUS test guideline about the autumn shoot, the flowers and the fruit edible rate only have grades and standard varieties, and there is no grade range for each grade, resulting in some difficulties to the accurate description of quantitative characteristics during litchi DUS testing. Therefore, the grade range and correlation of the nine quantitative characteristics were studied in order to provide technical basis for the accurate and objective description and observation of litchi quantitative characteristics, and for the correct determination of distinctness, uniformity and stability of litchi variety under testing.【Methods】Following the litchi DUS test guideline, nine quantitative characteristics were observed of the 74 litchi varieties in the Litchi Germplasm Resource Collection, College of Horticulture, South China Agricultural University. Samples were collected from the upper and middle periphery of the tree crown. 10 freshly green leaves and 10 annual branches were collected during the autumn shoot period to measure leaf length, leaf width, petiole length, branch thickness, internode length and leaf axis length. 10 flower spikes were collected in flowering stage to measure flower spike length and width. 10 mature fruits were collected in fruit ripening stage to measure the edible rate. The SPSS statistic software was used to test the normality of the data, and the R language was used for correlation analysis and mapping. For the quantitative characteristics conformed to be the normal distribution, the least signifi-cant difference method was used for grading. The average value of the measured data was taken as the center and the two sides were divided equally with the grade difference which was greater than or equal to 2 times LSD 0.05. For quantitative characteristics out of the normal distribution, the grade difference method was used for grading. Firstly, the grade difference of each quantitative characteristic was calculated by the ratio of the range and the grade number. Then the median of distribution range was used as the grade midpoint, and the range was divided according to the formula y= G±(1/2+n)x, (n = 0, 1, 2, 3, 4, G as the data median, x as the grade difference).【Results】The median and mean values of the 9 quantitative characteristics data were basically equal, indicating that the distribution of each characteristic of these 74 varieties were relatively regular without obvious extreme value, and the variation of the characteristics was mainly the result of natural selection. Among the 9 quantitative characteristics, the smallest coefficient of variation was the branch thickness, which was 12.35%, indicating that the expression status of the characteristic was similar. The largest coefficient of variation was the flower spike width, which was 33.09% , indicating that the genetic variation of this characteristic was rich. The variation range of the fruit edible rate was between 23.85% and 90.43%., which indicated that the tested varieties were representative. The results of K-S normality test showed that the bilateral significance (Sig.) of the leaf axis length, the branch thickness and the fruit edible rate was over 0.05, which was in line with normal distribution. The Sig. of the internode the length, the leaf length, the spike length and the spike width was less than 0.05, but because the absolute values of the skewness and the kurtosis were less than 1, and they could be treated as normal distribution. The Sig. of the leaf width and the petiole length was less than 0.05, as the absolute value of their kurtosis was over 1, and they were not conformed to be the normal distribution. Thus, in accordance with the normal distribution, the variance analysis of the leaf axis length, the branch thickness, the fruit edible rate, the internode length, the leaf length, the flower spike length and width was conducted, and the values of LSD0.05 were 0.53, 0.17, 0.056, 0.31, 081, 2.6, and 2.93 respectively. According to the above method, the seven quantitative characteristics were divided into 9 grades, 3 grades, 7grades, 7 grades, 9 grades, 7 grades, and 3 grades respectively. The leaf width and the petiole length that did not meet the normal distribution were divided into 9 and 3 grades using the range method. The correlation analysis of the 9 quantitative characteristics showed that there was a significant positive correlation between the spikelet length and the five characteristics including the internode length, the leaf axis length, the spikelet width, the leaf width and the leaf length, among them the correlation coefficient with the spikelet width was the highest (0.75). There was a significant positive correlation between the internode length and the four characteristics including the branch thickness, the flower spike width, the fruit edible rate and the petiole length. There was a significant correlation between the length of leaf axis and the four characteristics including the branch thickness, the flower spike width, the leaf width and the petiole length. The correlation coefficient between the branch thickness and the fruit edible rate was - 0.14. The correlation between the leaf width and the petiole length was very significant.【Conclusion】The results of the correlation analysis of 9 quantitative characteristics showed a very high correlation coefficient between the spikelet length and the spikelet width. Therefor, these two characteristics might be considered to merge into one characteristic in the test process to reduce the work of the DUS testing. As the quantitative characteristics are usually affected by the changes of environmental conditions and the data was only collected in Guangzhou, the classification results might not be completely suitable for the other ecological regions. Therefore, the classification scope needs to adjust according to the performance of standard varieties in different ecological regions.