Fine Individual Selection of Camellia pitardii in Pengshui County

Xiaomei TAN Rong ZHANG Yunfeng GUAN Xiangshun DENG Ya WANG

Abstract In order to select improved varieties， a study on the fine individual selection was carried out in Pengshui County， Chongqing Province. For 14 fine individuals with high and stable yield preliminarily screened， eight factors （fruit height， fruit diameter， pericarp thickness， fresh fruit weight， fresh seed weight， dry seed weight， dry seed yield and fresh seed yield） related to yield were selected to perform principal component analysis， and based on comprehensive scores and oil contents， five fine individuals （Longshe Town 7， 8， 12， 17 and Shaoqing Street 2） of Camellia pitardii were determined. These fine individuals have a better performance.

Key words Camellia pitardii; Fine individuals; Principal component analysis

Camellia oleifera Abel. is a woody oil tree species with high comprehensive utilization value in China[1-3]. It is one of the worldюs four major oil tree species including olive， oil palm and coconut. C. oleifera is often used as a garden plant because of its bright color and beautiful tree shape， which has high ornamental and economic value[4]. C. oleifera is a promising forestry industry in Chongqing， which can not only improve the mountain ecological environment， but also improve the economic level[5]. The cityюs C. oleifera area has exceeded 66 700 hm2[6]， and seven common C. oleifera varieties have been approved. However， C. pitardii is relatively backward in the breeding of improved varieties， and the research on the selection of superior C. pitardii individuals in Chongqing has not been reported. Therefore， it is necessary to carry out research on breeding of local C. pitardii.

General situation of selected area

Pengshui Autonomous County is located at 28°57′-29°51′N and 107°48′-108°36′ E， in the southeast of Chongqing. It belongs to the Wuling mountainous area and the lower reaches of the Wujiang River. This area has a midsubtropical warm monsoon climate， which is mild， and has sufficient rainfall and relatively less illumination. The annual average temperature is 17.50; the average annual rainfall is 1 104.20 mm; the annual sunshine hours are 1 035-1 553 h; and the frostfree period is over 312 d. The autonomous county has an altitude of 163-1 837 m， and the climate has large threedimensional differences. When the altitude is increased by 100 m， the average temperature is reduced by 0.46-0.55. The terrain is reduced from northwest to southeast， with a denudation landform， of which 13% is hilly valley， 53% is low mountains， and 34% is middle mountains. The soil is dominated by yellow soil， yellow brown soil and purple soil[7-8 ].

Materials and Methods

Experimental materials

A total of 31 C. pitardii plants were investigated， 14 of which were originally distributed in Longshe Town， Pengshui County， and now planted in Gaoping Village， Houping Township， at the interface of Pengshui County and Wulong County after ex situ conservation. Other 17 plants were distributed in Shaoqing Street， Hanjia Street， Dianshui Street and Yandong Township of Pengshui Autonomous County.

Experimental methods

Through three years of consecutive visits and field observation， excellent plants with good growth and full buds free of pests and diseases were screened out， marked with brightly colored spray paint， and further investigated to determine 22 tree traits such as their elevation， latitude and longitude， tree shape， tree gesture， tree height， clear bole height， ground diameter， crown width and leaf area. All fruit was weighed after harvesting， and 30 individuals were randomly selected and timely determined for fruit and seed traits including fruit height， fruit diameter， peel thickness， seed number， fresh seed weight and dry seed weight.

Selection and quantification of indicators

The main purpose of the C. pitardii selection was to screen out individuals with high yield， stable yield and excellent comprehensive performance. After three years of consecutive yield estimation， 14 individuals （specific tree conditions of which were shown in Table 1） with high and stable yield were screened out of the 31 individuals for comprehensive trait analysis. From the abovementioned indices， such nine factors related to the fruit yield of C. pitardii as fruit height， fruit diameter， peel thickness， fresh fruit weight， fresh seed weight， dry seed weight， dry seed yield， fresh seed yield， and seed oil content were selected （as shown in Table 2）. The values of the selected indices were all actually measured or calculated values. The seed oil content of the selected indicators should be analyzed separately as a rigid index for selecting good individual plants.

Data analysis

Data statistics and analysis were performed using SPSS 20.0 statistical software[9].

Results and Analysis

Analysis on discrete degree of various index factors of C. pitardii

The variations of various index factors in C. pitardii are shown in Table 3. It could be seen from Table 3 that the range values of the eight yieldrelated index factors （fruit height， fruit diameter， peel thickness， fresh fruit weight， fresh seed weight， dry seed weight， dry seed yield and fresh seed yield） were all larger. Among them， the fresh fruit weight had the largest variation， from 356.8 to 1 132 g; and the dry seed yield had the smallest variation， from 0.12% to 0.33%. The coefficients of variation of the various index factors ranked as peel thickness > dry seed weight > fresh seed weight > dry seed yield > fresh fruit weight > fresh seed yield > fruit height > fruit diameter. The coefficient of variation of the peel thickness was the largest， of 67.77%; and the coefficient of variation of the fruit diameter was the smallest， of 18.66%. The variations of various index factor of C. pitardii were all relatively large， which provides a large space for further breeding of C. pitardii.

Principal component analysis of various index factors of C. pitardii

Selection of principal components

Principal component analysis was performed on the excellent 14 C. pitardii individual plants initially screened using the eight index factors related to yield. It could be seen from Table 4 that the KMO value was 0.554 and the P value of Bartlett sphericity test was smaller than 0.05， which is in accordance with the principal component analysis conditions. In the principal component analysis， the selection of principal components was mainly based on the eigenvalues and the accumulated variance contribution rate. As can be seen from Table 5， the principal component composition information was mainly concentrated in the first three principal components. The eigenvalue of the first principal component was 4.106， and the variance contributing rate was 51.321%， which represented its proportion in all the trait information; the eigenvalue of the second principal component was 2.375， and the variance contribution rate was 29.692%， so it was the important principal component second only to the first principal component; and the third principal component had an eigenvalue of 1.356， and the variance contribution rate was 16.954%， which was relatively lower. The accumulated contribution rate of the first three principal components was 97.967%， which can reflect the overall information of the selected superior plants. Therefore， the selected three principal components were used as comprehensive indices for the final superior plant selection.

Establishment of functions

The loading values of the three principal components extracted by principal component analysis are shown in Table 6. It could be seen from Table 6 that such five index variables as fruit height， fruit diameter， peel thickness， dry seed yield and fresh seed yield had higher loads on the first principal component， indicating that the first principal component basically reflected these indices; the fresh fruit weight， fresh seed weight and dry seed weight had higher loads on the second principal component， indicating that the second principal component basically reflected the information of the three indices; and the third principal component had no higher loading value， and was an auxiliary component.

The coefficients corresponding to each of the three principal components can be obtained by extracting the square roots of the quotients of the factor loading values in Table 6 and the eigenvalues corresponding to each principal component. The results are shown in Table 7. From Table 7， we can conclude that the expressions of the three principal components were：

Z1=-0.38X1-0.42X2-0.44X3-0.08X4+0.25X5+0.29X6+0.41X7+0.39X8

Z2=0.18X1+0.17X2+0.16X3+0.58X4+0.55X5+0.52X6-0.05X7-0.03X8

Z3=0.46X1+0.34X2+0.27X3-0.36X4+0.07X5+0.04X6+0.45X7+0.51X8

The variable X here should be standardized variable.

Comprehensive evaluation

In the process of selecting superior plants， it is difficult to explain the comprehensive traits of a selected individual using a single index or a single component. Therefore， it is necessary to construct a comprehensive evaluation index to determine the final selected superior plants. Taking the proportion of the variance contribution rate corresponding to each principal component in the accumulated contribution ratio of the three principal components as a weight， the principal component comprehensive evaluation mathematical model was calculated as：

The larger the comprehensive score value， the better the performance of the selected plant. According to the comprehensive evaluation model， the comprehensive scores of the superior individuals of C. pitardii were calculated and ranked. The results are shown in Table 8. It could be seen from Table 8 that the top scores of the top 9 plants were positive， and the overall performance was higher than the average; and the comprehensive scores of latter superior plants were negative， and the overall performance was lower than the average.

Selection results of superior trees

The comprehensive scores obtained by principal component analysis and the rigid index seed oil content were the criteria for selecting superior plants， and the combination of the two determined the superiority of the selected plants. Firstly， the comprehensive scores of 14 selected superior plants were subjected to hierarchical cluster analysis using the clustering function of SPSS analysis software， and they were divided into three levels， which were designated as I， II and III， respectively. The first two levels initially determined the superior plants to be selected in the next step， as shown in Table 9. It could be seen from Table 9 that there were 11 selected superior plants with the comprehensive score ranging from -0.28 to 1.39. The numberings were Longshe Town 8， 15， 12， 7， 6， 17， 10， 9， 5， 11 and Shaoqing Street 2， respectively. Finally， the superior plants with a seed oil content larger than the average of the selected superior plants were selected. The results are shown in Table 10. It could be seen from Table 10 that the average oil content of the 11 superior plants was 18.0， and there were six plants having a seed oil content larger than 18.0. However， the comprehensive score of Longshe Town 11 was -0.28， which was lower than the average level， so it was rejected. Therefore， there were five superior C. pitardii individuals with better comprehensive score and seed oil content， namely： Longshe Town 7， 8， 12， 17 and Shaoqing Street 2.

Discussion and Conclusions

The selection of superior C. pitardii individuals can provide superior germplasm resources for C. pitardii breeding in Chongqing. In the process of selecting superior individuals， it is often necessary to investigate and observe a large number of indices. Although multivariate large samples provide rich information for superior individual selection， they increase the difficulty of data processing. And there is certain correlation between variables， which leads to the overlapping of information， which increases the complexity of problem analysis. Principal component analysis aims to transform multiple indices into a few related comprehensive indices according to the idea of dimensionality reduction under the premise of losing less information. Therefore， principal component analysis as a method for selecting superior C. pitardii individuals is more simple and scientific[10-13].

Through principal component analysis， eight yieldrelated indices of the 14 superior C. pitardii individuals preliminarily screened were firstly transformed into three principal components， and three functional expressions of principal components were established. Then， a comprehensive evaluation model was constructed with the percentage weighting coefficient of the variance contribution rates. Finally， five superior trees were determined according to the comprehensive scores and seed oil contents， namely： Longshe Town 7， 8， 12， 17 and Shaoqing Street 2， which can provide good materials for breeding of C. pitardii in Chongqing City.

Xu et al.[14]reported that the seed oil contents and dry seed rates in finally determined superior C. pitardii individuals in Tengchong were 50.13%-53.65% and 12.01%-17.38%， respectively. In contrast， the C. pitardii selected by this project had slightly lower seed oil contents （18.1%-28.2%）， but the dry seed yields （23%-32%） were significantly higher. Although the seed oil contents of the C. pitardii detected by the project were generally low， there were also more individuals with higher single or multiple traits， which can be collected and preserved as good germplasm resources， providing good parental materials for later hybrid breeding and good variety breeding. For instance， Longshe Town 15， which had higher yield， dry seed yield and fresh seed yield， with a comprehensive score ranking second， but lower oil content， can be used as a highyield parent material for variety hybridization to aggregate good traits， to thereby improve it for application in production.

References

[1]KONG WJ， LIU XL， YAO XH， et al. Research on photosynthetic characteristics of four oiltea （Camellia） species[J]. Journal of Southwest University： Natural Science Edition， 2013， 35（1）：  16-22. （in Chinese）

[2]PENG SF， CHEN YZ， WANG XN， et al. Research progress and development strategies of breeding of oiltea Camellia[J]. Hunan Forestry Science & Technology， 2010， 37（6）： 32-38. （in Chinese）

[3]DING SJ. Making innovation to accelerate the development of Chinaюs oiltea Camellia industry[J]. China Oils and Fats， 2009， 34（3）： 1-5. （in Chinese）

[4]ZHOU J， ZHOU JP， TAN HY. Main components analysis of Camellia chekiangoleosa Hu fruit[J]. Modern Food Science & Technology， 2010， 26（1）： 109-113. （in Chinese）

[5]PENG X， LI XZ， YAN Q， et al. Comparative study on the early growth performance of fine varieties of Camellia oleifera in Chongqing area[J]. Forestry Science & Technology， 2013， 38（2）： 1-5. （in Chinese）

[6]LOU LH， PENG X， LI XZ， et al. Effects of different nursery substrates on the growth of grafting plants of oil Camellia[J]. Journal of Sichuan Forestry Science and Technology， 2012， 33（3）： 75-77. （in Chinese）

[7]WU LL. The problems and counterplans of agriculture specialization development in Pengshui County[D]. Chongqing： Southwest University， 2012. （in Chinese）

[8]CHU QN. Research on the relationship climate and soil with tobacco quality in tobaccogrowing areas in Chongqing[D]. Chongqing： Southwest University， 2012. （in Chinese）

[9]CHEN SK. SPSS statistical analysis from getting started to proficient[M]. Beijing： Tsinghua University Press， 2013： 358-361. （in Chinese）

[10]HUANG Y， LANG NJ， XU DB， et al. A comprehensive selection of excellent Camellia oleifera with multiple indices from Yunnan Province[J]. Journal of West China Forestry Science， 2011（3）： 43-47. （in Chinese）

[11]ZHANG ZJ， YANG SX， ZENG YJ， et al. Variation within clones and families and superior individual selection in different cultivars of Camellia oleifera ‘Ruanzhi’[J]. Journal of Beijing Forestry University， 2016， 38（10）： 59-68. （in Chinese）

[12]YE HL， WANG LC. Multiindex comprehensive selection of wild Camellia saluenensis Stapf ex Bean[J]. Journal of Central South University of Forestry & Technology， 2016（6）： 66-69. （in Chinese）

[13]CHEN MH. Superior individual selection of Camellia oleifera in Yanping Disctrict[J]. Journal of Fujian Forestry Science and Technology， 2014（1）： 97-102. （in Chinese）

[14]XU ZY， XIE Y， XIN CL， et al. Fine individual selection of Camellia oleifera[J]. Forest Inventory and Planning， 2015， 40（2）： 96-101. （in Chinese）