Correlation Analysis and Principal Component Analysis of Eight Agronomic Traits of Autumn Broad Beans (Vicia faba L.)

Li LI Zhenghuang WAN Changyan LIU Liangjun LIU Hongwei CHEN

Abstract ［Objectives］ This study was conducted to give play to the yieldincreasing potential of new broad bean varieties.

［Methods］ The correlation analysis and principal component analysis of eight main agronomic traits of 15 autumnsown broad bean varieties in 2017-2018 were carried out.

［Results］ There were extremely complicated correlations between the agronomic traits. Among the various agronomic traits, the number of pods per plant showed a significant negative correlation from the number of seeds per pod (P<0.05), and significant negative correlations with pod length and 100grain weight (P<0.01). The number of seeds per pod was significantly positively correlated with pod length (P<0.01). The 100grain weight of broad beans was significantly positively correlated with the pod length (P<0.01), and was significantly negatively correlated with the number of pods per plant (P<0.01). Three principal components were established by principal component analysis, and the contribution rate of the first principal component was the highest, which was 40.18%. The contribution rates of the second principal component and the third principal component to the total variance were 24.08% and 13.55%, respectively. The score equations of the principal components were derived.

［Conclusions］ This study provides a theoretical reference for the application and promotion of new varieties of autumn broad beans.

Key words Vicia faba L.; Agronomic trait; Correlation analysis; Principal component analysis

Received: May 29, 2019Accepted: August 27, 2019

Supported by Hubei Agricultural Science and Technology Innovation Center Fund (2016620000001006); Hubei Agricultural Germplasm Resource Sharing Platform (PTJX2015000008); National Modern Agriculture Industrial Technology System Construction Project (CARS09).

Li LI (1974-), female, P. R. China, associate researcher, master, devoted to research about breeding and cultivation of edible beans.

*Corresponding author. Email: zhwan168@163.com; hwchen25@126.com.

Broad bean (Vicia faba L.), also known as Hudou, Fodou, Chuandou, Wodou and Luohandou, is generally believed to originate from southwest Asia and northern Africa. It was introduced to China from Europe in the 1st century AD. It is grown in most parts of the country. The planting area of broad beans in Hubei Province is about 35 000 hm2, mainly concentrated in Jingzhou, Shiyan, Xianning and other places. It is mainly grown in autumn and winter［1-2］.

Broad bean is an important food, vegetable, nonstaple food, feed, green manure and landcultivation crop［3-5］. Due to its wide range of uses, large production potential and broad market prospects, the planting area has expanded year by year. Good varieties are the dominant factor for achieving high yield and stable yield of broad beans, and are an important support for ensuring Chinas grain harvest and safety, as well as one of the effective means to promote industrial development［6-7］. In order to give full play to the yieldincreasing potential of new broad bean varieties, we identified the high yield, stability, adaptability, resistance and other important characteristics of domestic broad bean varieties (combinations) in Hubei Province, explored the interrelationship between the main agronomic traits of broad beans, and carried out principal component analysis to comprehensively evaluate and screen multiple traits of multiple varieties, so as to provide a theoretical reference for the application and promotion of new varieties.

Materials and Methods

General situation of test site

The test site is located in the scientific research base of Institute of Food Crops, Hubei Academy of Agricultural Sciences, with an altitude of 20 m. The test site is flat and has uniform fertility. The soil is clay loam with a pH value of 6.7, and the organic matter content is about 1.8%. The previous crop was maize. Plowing was performed with a tractor before planting, and the residual weeds were removed. The prepared land was basically flat, fine and loose.

Test materials

A total of 15 broad bean varieties were tested, all of which are domestic excellent varieties. The names and sources are shown in Table 1.

Test design

The field trial adopted random block design. Each treatment was done in 3 replicates. Each plot had 4 rows, which had a row length of 5 m and row spacing of 0.4 m. The hole spacing was 0.48 m, and each row included 30 holes. Each plot had an area of 10 m2, surrounded by guard rows. Conventional field management was adopted.

Data investigation

From each plot, samples were collected from the third plant of the middle row continuously, and a total of 10 plants were collected for field observation and laboratory test. According to the description of related traits in the "Descriptors and Data Standard for Faba Bean", field investigation and laboratory test were conducted on 8 types of traits. The identification items and standards are shown in Table 2.

Data processing and statistical analysis

The basic data processing was performed by Excel (Microsoft office, 2007) data software, and principal component analysis and correlation analysis were performed using SPSS data processing software. Eight following parameters were subjected to principal component analysis: growth period (X1), plant height (X2), number of branches per plant (X3), number of pods per plant (X4), number of seeds per pod (X5), pod length (X6), 100seed weight (X7) and plot yield (X8). In order to avoid the influences of differences in the dimension and magnitude between variables on the calculation results, and to ensure its objectivity and scientificity, the original data matrix was standardized, parameters including correlation coefficient, eigenvalue, eigenvector, contribution rate and accumulating contribution rate were calculated, and principal components were extracted.

Results and Analysis

Morphological characteristics and biological characteristics of broad bean varieties

The morphological and biological characteristics of various broad bean varieties are shown in Table 3. The growth period of broad beans is the most important indicator to judge the sensitivity to light and temperature. The growth periods of the tested broad bean varieties were about 175-186 d, with an average of 180.5 d and a range of 11 d. According to the analysis, the growth periods of broad bean varieties from Yunnan and Sichuan were mostly short, while the growth periods of broad beans from Jiangsu and Hubei were generally long, indicating that due to different natural conditions, planting systems, breeding preferences, etc., significant regional differences between varieties from different sources have emerged. The plant heights of broad bean varieties ranged from 74.1 to 106.0 cm. The differences in pod length between varieties were large. The longest pod length was 12.2 cm, the shortest was 6.5 cm, and the range reached 5.7 cm. The numbers of pods per plant were between 8.3 and 23.2, with a range of 14.9. The 100seed weights of the tested broad beans showed a largest value at 201.0 g and a smallest value at 68 g, and the differences were quite large. The average plot yield of broad beans was 2.4 g, and Yucandou No. 2 and Yucandou No. 1 from Sichuan had a higher yield among the various varieties, followed by Yundou 459 and Feng 01105.

Correlation analysis on main quantitative traits of broad bean varieties

From Table 4, it can be seen that the differences in plant height, pod length, number of pods per plant and 100seed weight between the 15 tested broad bean varieties were all large, and there were extremely complicated correlations between the agronomic traits. Among the various agronomic traits, the plant height had an extremely significant difference from the number of branches per plant (P<0.01), but its correlations with other quantitative traits were small. The growth period of broad beans had a significant positive correlation with 100seed weight (P<0.05), and the correlations with other six traits were not significant. The number of pods per plant showed a significant negative correlation from the number of seeds per pod (P<0.05), and a significant negative correlation with the pod length and 100grain weight (P<0.01). The number of seeds per pod was significantly positively correlated with the pod length (P<0.01), indicating that the longer the pods of broad beans, the more the seeds per pod. The 100grain weight of broad beans was significantly positively correlated with the pod length (P<0.01), and was significantly negatively correlated with the number of pods per plant (P<0.01), indicating that increasing the pod length of broad beans and reducing the number of pods per plant are the key to increasing the 100seed weight of broad beans.

Principal component analysis of major quantitative traits in broad bean varieties

Principal component analysis was carried out on eight agronomic traits of tested germplasm resources. The value of each characteristic root represents the genetic variance of each comprehensive index, and the cumulative percentage of each characteristic root represents the percentage of contribution of each relevant comprehensive index to the total genetic variance. The calculated contribution rates and cumulative contribution rate of eigenvalues were calculated. (Table 5), and the principal components were obtained according to the principle of cumulative contribution rate≥75.34%. Accordingly, three principal components were extracted. The contribution rates of the first, second and third principal components to the total variance were 40.18%, 24.08% and 13.55%, respectively, and the sum of the three reached 77.81%, that is, the first three principal components can reflect 77.81% of the information provided by all indicators. Therefore, it is reliable to use principal component analysis to investigate the correlations between various traits of tested broad beans.

By analyzing the scores of the three principal components, it could be known that the correlations of the first principal component with the number of pods per plant, pod length and 100seed weight were large, and the weight coefficients of the number of pods per plant (X4) and pod length (X6) were larger, reaching -0.51 and 0.51, respectively. The correlations of the second principal component with the plant height (X2), number of branches per plant (X3) and plot yield (X8) were 0.62, 0.62 and 0.46, respectively, which constituted the main source of variation. The correlations of the third principal component with the growth period (X1), number of seeds per pod (X5) and 100seed weight (X7) were 0.52, -0.63, and 0.41, respectively, indicating that the resources with a longer growth period, fewer pods per plant and a higher 100seed weight scored higher in the third principal component.

The three principal components were represented by y1, y2 and y3, respectively, and the score equations of various principal components are as follows:

The first principal component:

y1=0.35x1-0.07x2+0.05x3-0.51x4+0.37x5+0.51x6+0.44x7+0.15x8

The second principal component:

y2=0.01x1+0.62x2+0.62x3+0.14x4-0.03x5-0.003x6+0.06x7+0.46x8

The third principal component:

y3=0.52x1+0.18x2-0.31x3-0.003x4-0.63x5-0.22x6+0.41x7+0.07x8

Conclusions and Discussion

Broad bean varieties are important natural resources formed by natural selection and artificial selection. They contain rich genetic variations caused by natural and manmade factors. Therefore, to introduce and collect a large number of variety resources and to understand and master their characteristics through screening and evaluation are of great significance for the rational development and full utilization of bred varieties［8-10］.

The tested 15 wild bean variety resources from Yunnan, Jiangsu, Sichuan and Hubei provinces all could well adapt to the ecological conditions of Hubei, and exhibited characteristics and differences in such eight agronomic traits as the growth period, plant height, number of branches per plant, number of pods per plant, number of seeds per pod, pod length, 100seed weight and plot yield inherent in each variety. The ranges in pod length, number of pods per plant and 100seed weight were 5.7 cm, 14.9 pods and 133 g, respectively, and the differences were very obvious. The higher population yields were from Yucan No. 2 and Yucan No. 1 from Sichuan, followed by Yundou 459, Feng 01105 and Tongcanxian No. 6 from Yunnan and Jiangsu.

There were complicated correlations between the agronomic traits of the tested varieties. Among the various agronomic traits, the number of pods per plant was in significant negative correlations with the number of seeds per pod, pod length and 100seed weight (P<0.01), indicating that with the increase of the number of pods per plant, the number of seeds per pod, pod length and 100seed weight were all on the decease. The pod length of broad beans had significant positive correlations with the number of seeds per pod and seed weight (P<0.01), indicating that the pod length was the key to increasing seed number and seed weight of broad beans. The growth period was in a significant correlation with 100seed weight (P<0.05), and the correlations with other quantitative traits were small. The plant height of broad beans was significantly positively correlated with the number of branches per plant (P<0.01).

Through the principal component analysis, three principal components were extracted, and the first principal component contributed 40.18% to the total variance, and had larger correlations with the number of pods per plant, pod length and 100seed weight. The contribution rate of the second principal component to the total variance was 24.08%, and it was highly correlated with the plant height, number of branches per plant and plot yield, with the correlation coefficients of 0.62, 0.62 and 0.46, respectively. The third principal component had large correlations with the growth period, number of seeds per pod and 100seed weight, and its contribution rate to the total variance was 13.55%. The cumulative percentage of the above three principal components reached 77.81%, indicating that the traits of the 15 broad bean varieties were reliable from principal component analysis, and the three principal components reflected 77.81% of the information of the original data.

References

［1］ LYU JQ, YANG M, LI Y, et al. Advances in the breeding of Vicia faba L. in China［J］. Journal of Anhui Agricultural Sciences, 2012, 40(3): 1403-1406. (in Chinese)

［2］ GU CL, YIN HP, et al. The review of produce and develop trend of faba bean in China［J］. Journal of Yunnan Agricultural University, 2005, 20(5): 671-675. (in Chinese)

［3］ Li L, LIU CY, CHEN HW, et al. Identification of faba bean (Vicia faba L.) germplasm resources in Hubei Province and evaluation on their main agronomic traits［J］. Agricultural Basic Science and Technology, 2017, 18(12): 2219-2222.

［4］ DU CZ, ZHANG JJ, ZENG XQ, et al. Effects of different sowing dates on agronomic characters and yield and of broad bean in Chongqing［J］. Bulletin of Agricultural Science and Technology, 2010, 12: 86-89. (in Chinese)

［5］ YANG M, LI Y, ZHENG JM, et al. Effect of different density and fertilizer levels on agronomic traits and yield of faba bean［J］. Chinese Agricultural Science Bulletin, 2012, 28(24): 133-137. (in Chinese)

［6］ LI L, WANG MC, ZHANG CY, et al. Breeding of a new faba bean variety Ecandou No.1 and its cultivation techniques［J］. Hubei Agricultural Sciences, 2016, 55(16): 4100-4102. (in Chinese)

［7］ LYU CY, LIAO FL, CHEN HW, et al. Identification and evaluation of yield traits of 41 Vicia faba germplasm resources from African areas and Hubei, China［J］. Journal of Southern Agriculture, 2018, 49(12): 2356-2363. (in Chinese)

［8］ LIAO YM, WANG XJ, CHEN MF, et al. Genetic variation, correlation and principal component analysis on major agronomic traits of vegetable faba bean［J］. Journal of Hebei Agricultural Sciences, 2010, 14(10): 95-97. (in Chinese)

［9］ YU L, ZHANG SL, LI QC, et al. Principal component analysis of main agronomic traits of autumn broad beans［J］. Chinese Agricultural Science Bulletin, 2015, 31(27): 108-113. (in Chinese)

［10］ LI HM, DING CH, ZHANG GE, et al. Preliminary report on the variety comparative test on broad beans as a vegetable［J］. Journal of Zhejiang Agricultural Sciences, 2012, (1): 50-51. (in Chinese)