【作者】 李艳艳；

【导师】 丰震； 赵兰勇；

【作者基本信息】 山东农业大学 ， 园林植物与观赏园艺， 2008， 硕士

【摘要】 本研究是在对我国玫瑰(Rosa rugosa Thunb.)资源进行全面调查的基础上,对玫瑰花数量性状的遗传参数和选择效率进行了初步研究,并对玫瑰花单株产量进行了灰色分析、Kriging插值和选择指数的研究,在此基础上进行了BP神经网络模型和AMMI模型分析,并对玫瑰花期等性状进行了遗传变异分析和多变量时间序列CAR模型分析。同时为进行新品种的选育,选取了不同的玫瑰品种用秋水仙素进行了多倍体育种研究;并在观察调查的基础上选取了不同的玫瑰品种为亲本进行了杂交授粉的研究,这将进一步丰富玫瑰种质资源。主要结论如下:1、对20个玫瑰品种的鲜花产量及13个数量性状进行了遗传相关、相关遗传力及遗传系数通径分析等。结果表明:(1)单株花蕾数、分枝数、第1个花瓣长、第2个花瓣长、花蕾长与单株花蕾产量的遗传相关显著;(2)单株花蕾数、第2个花瓣长、分枝数和花蕾长的相关遗传力与单株花蕾产量的遗传力接近,相关选择的效率接近对单株花蕾产量的直接选择的效率;遗传系数通径分析结果进一步表明第2个花瓣长、花蕾长和花蕾宽对单株花蕾产量的直接作用都比较大;(3)根据决策系数,花蕾长、单株花蕾数、第2个花瓣长、分枝数、单蕾花瓣鲜重对单株产量有促进作用。2、为了探讨玫瑰高产品种的选育理论与技术,对20个玫瑰品种的15个数量性状与单株产花量进行了灰色系统分析、Kriging插值模拟、综合选择指数分析,结果表明:单株花数、分枝数、花蕾宽和单花鲜重4个性状与单株产花量的灰色关联度较大(>0.5)。根据单株产花量的Kriging插值模拟结果,单株产花量随单株花数、分枝数的增加而增加,任何单一因素的变化均不能保证主要目标性状的提高。因此,提高玫瑰单株产花量的选育方法应采用多性状综合选择法,选用单株花数、分枝数、花蕾宽和单花鲜重4个性状,以关联度为权重建立了单株产花量综合选择指数模型:I=0.3x1-318.6x2+670.1x4+6.3x8,指数遗传力=0.8014,综合育种值选择进展ΔH=245.8811,该研究结果对今后玫瑰高产花量品种的选育具有指导意义。3、选用13个玫瑰品种,连续两年测定了各品种的单株产花量,应用AMMI(Additive main effects and multiplicative interaction,又称为主效可加互作可乘)模型对单株产花量的基因型、环境和基因型与环境(G×E)的互作进行了探讨。结果表明:基因型、环境及G×E互作的平方和分别占总平方和的65.610%、12.352%、22.038%,均达极显著水平,而误差仅占2.75×10-17%,参试品种的单株产花量主要在500～1500g之间;AMMI双标和排序图表明品种‘紫云’、‘玉盘’、‘唐紫’、‘唐粉’、‘紫枝玫瑰’、‘朱龙游空’与2006年的环境的互作为正,而与07年的环境互作为负;‘赛西子’、‘唐红’、‘西子’、‘紫芙蓉’、‘朱紫双辉’、‘紫雁’、‘香刺果’与07年的环境互作为正,与2006年的环境的互作为负。AMMI品种适应性分析显示‘朱龙游空’、‘唐紫’和‘赛西子’具有最佳适应性。AMMI模型很好的解释了玫瑰品种产量性状的基因型效应、环境效应和G×E互作效应,根据分析结果可以得出以下结论:单株产花量高且稳定的品种有‘西子’、‘紫芙蓉’和‘赛西子’(1200～1800g),相对稳定的品种有‘玉盘’、‘唐粉’、‘紫枝玫瑰’、‘紫云’、‘紫雁’和‘朱紫双辉’(800～1150g),高产但较不稳定的有‘唐紫’和‘朱龙游空’(1700～2600g),产量低也不稳定的是‘唐红’和‘香刺果’(500～600g)。4、对20个玫瑰品种的8个数量性状进行方差分析的基础上,对产花量性状进行了重复力的分析,结果表明:各性状间均达到极显著差异,各品种单株产量的重复力都较高且均在0.85以上,为今后玫瑰的遗传育种和资源研究提供了理论依据。5、对20个玫瑰品种的所有数量性状进行反向传播神经网络(BP)分析、逐步回归,选择与单株产花量相关性比较大的单株花蕾数进行一元线性回归和一元非线性模型的二次曲线分析来预测单株产花量。结果表明:BP神经网络对所有品种的单株产花量的预测拟合结果的相对误差的绝对值最小(均小于10%),最高的达9.060%;BP神经网络的预测单株产花量的曲线与实测值的结果最接近,二者几乎达到了重叠,初步说明BP神经网络方法是一种可用于玫瑰产量预测的模型的新模拟方法。利用BP神经网络模型,可以对玫瑰单株产花量进行准确的预测,研究结果对今后玫瑰高产花量的预测具有指导意义。6、本文首次对13个主要栽培玫瑰品种的花期、花径、花朵数和花瓣数等与观赏特性有关的性状进行了观测和遗传变异分析,结果表明:玫瑰的花期、花径、花朵数和花瓣数等性状的差异性检验达到了极显著,而且各性状的遗传力(h2)都比较高(0.5779-0.7462),遗传变异系数0.0315-0.5235;花期及其构成单元现蕾期、初花期、盛花期和凋谢期的变异模式非常丰富,花期与花径、花瓣数、花朵数之间组合方式非常多样;单株花朵数多、花期长的有‘朱龙游空’、‘唐紫’、‘唐红’和‘西子’,花瓣数多、花期长的有‘朱龙游空’、‘紫枝玫瑰’、‘唐紫’、‘香刺果’等,‘紫枝玫瑰’花期早、花期长且多次开花;最后建立了花期的CAR(n)模型,讨论了玫瑰花期等观赏特性的育种策略,也为玫瑰杂交育种亲本交配设计提供了理论依据,这将有助于进一步提高玫瑰的观赏价值。7、本试验对2n花粉诱导和杂交育种技术进行了研究,通过对花粉母细胞减数分裂过程的观察,可以看出在花瓣尚未露出的花蕾(此时花瓣呈乳黄色,花药乳白色)的时期是进行2n花粉诱导的最好时期;杂交育种的结果表明以不同玫瑰品种为亲本的杂交坐果率有很大差异,为今后玫瑰的杂交亲本选择提供参考。更多还原

【Abstract】 The article studied the comprehensive investigation in Chinese rose (Rosa rugosa) and preliminary research on genetic parameters and selection efficiency of quantitative traits of fresh floral bud in Rosa rugosa. The analysis of grey system, Kriging interpolation and integration selection index were employed to investigate the flower number/plant. Based on it the analysis of Back Propagation (BP) neural network and AMMI model were done on the floral yield. There were considerable differences in drought-resistance in rugosa rose cultivars by the preliminary study on drought-resistance in Rugosa Rose cultivars and provenances. The genetic variation and CAR model of multivariable time Series on flowering period etc. in Rugosa Rose was analyzed. The different cultivars were selected to do polyploid induction breeding of new cultivars by colchicine and hybrid pollination. This will enrich the germplasm resource in Rugosa rose. The main results are as following:1.The floral bud yield/plant (y) and 13 quantitative traits of 20 rugosa rose cultivars were investigated. Analysis of genetic correlation, correlative heritability (hxy) and path analysis were carried out. The results indicated: (1) the number of floral bud/plant (x8), the number of branches (x13), the length of the first petal (x3), the length of the second petal (x5), the length of floral bud (x1) had significant genetic correlation relationship with y; (2) hxy of x8, x5, x13 and x1 related to y were higher respectively, and which were close to heritability (h2) of y. The efficiency of correlation selection of these traits would not be much lower than direct selection of y; the result of path analysis of genetic coefficient showed further that the length of the second petal (x5), the length and width of floral bud have higher direct effect of the floral bud yield/plant. (3) x1, x8, x5,x13, the fresh weight of pedals/floral bud (x9) played an important positive part in improving y according to decision coefficient.2.Seeking for the theory and techniques of selection breeding of high flower yield rugosa rose cultivars, the analysis of grey system, Kriging interpolation and integration selection index were employed to investigate the relationship between the flower yield per plant and 15 quantitative traits of 20 rugosa rose cultivars. The result showed that: The grey relational grade (GRG) of the flower number/plant, the number of branch, the width of floral bud and the weight of single flower to the flower yield/plant were larger (> 0.5). Kriging interpolation simulation was applied to analyze the flower yield/plant. It was found that the value of target trait went up with increase of the number of flower/plant and the number of branch. Moreover, the indirect selection of either trait couldn’t get better improvement of flower yield/plant. It’s necessary to improve flower yield/plant by multi-trait selection. The integration selection index equation of flower yield/plant was established with the characters flower number/plant, the number of branch, the width of floral bud and the weight of single flower. I =0.3x1 -318.6 x2 +670.1x4 +6.3x8. Index heritability = 0.8014, selective response of the integration breeding value = 245.8811. This will provide theoretic base for genetic breeding of rugosa rose.3. 13 Rugosa Rose cultivars were employed in this experiment. The flower yield/plant during continuous two years was measured and analyzed by AMMI (Additive main effects and multiplicative interaction) model. The article discussed the genotype, environment and interactive effects of genotype by environment (G×E) of flower yield/plant. The results showed that the proportions of the sum of squares of genotype, environment and G×E interactive effects to total sum of squares was 65.610%、12.352%、22.038% respectively, the error only 2.75×10-17%. And there were significant effects in genotype, environment and G×E interaction. The flower yield/plant in the 13 cultivars was mainly between 500-1000g. AMMI plots and taxis plots showed that there were positive interaction effects of cultivars R. rugosa‘Purple Cloud’, R. rugosa‘Jade Plate’, R. rugosa‘Tang Purple’, R. rugosa‘Tang Pink’, R. rugosa‘Purple Branch’, R. rugosa‘Puce Dragon’by environment in 2006 and negative by environment in 2007, positive interaction effects of cultivars R. rugosa‘Saixizi’, R. rugosa‘Tang Red’, R. rugosa‘Xizi’, R. rugosa‘Zifurong’, R. rugosa‘Zhuzishuanghui’, R. rugosa‘Purple Goose’, R. rugosa‘Xiangciguo’by environment in 2007 and negative by environment in 2006. The adaptability analysis revealed that R. rugosa‘Puce Dragon’, R. rugosa‘Tang Purple’, R. rugosa‘Saixizi’ were optimal adaptation to environment. AMMI model explained clearly the genotype, environment and G×E interactions of yield trait in rugosa rose cultivars. We can reach the conclusion that: The most stable and high output cultivars were R. rugosa‘Xizi’, R. rugosa‘Zifurong’and R. rugosa‘Saixizi(’1200～1800g). The relative stable ones were R. rugosa‘Jade Plate’, R. rugosa‘Tang Pink’, R. rugosa‘Purple Branch’, R. rugosa‘Purple Cloud’, R. rugosa‘Purple Goose’and R. rugosa‘Zhuzishuanghui’(800～1150g). The high yield but less unstable cultivars were R. rugosa‘Tang Purple’and R. rugosa‘Puce Dragon’(1700～2600g). The lower yield and unstable cultivars were R. rugosa‘Tang Red’and R. rugosa‘Xiangciguo’(500～600g).4. Based on the ANOVA of 8 quantitative traits in 20 Rugosa rose cultivars, the analysis of repeatability was investigated in the research. The result showed that there was significant difference in each trait. The repeatability of the yield per plant in every cultivar was great (over 0.85). This will provide theoretic base for genetic resource research and breeding of Rugosa rose.5. All the quantitative traits of 20 rose cultivars were presented to the analysis of Back Propagation (BP) neural network and stepwise regression. The flower number/plant which was more correlation with the flower yield/plant was selected to do the analysis of one-variable linear regression and conic of one-variable nonlinear to predict the flower yield/plant. The result showed that the absolute values of relative error of the prediction in flower yield/plant of all cultivars were the smallest by the analysis of Back Propagation (BP) neural network (<10%), the highest was 9.060%. The curve of Back Propagation (BP) neural network in flower yield/plant was near to the actual values, they were almost overlapped. This suggested that Back Propagation (BP) neural network was a new simulation method in the application to prediction in the yield. The flower yield/plant was accurately predicted by Back Propagation (BP) neural network. It will provide guiding significance for high flower yield.6. The study presented, for the first time, the observation, analysis of genetic variation and CAR(n) model on flowering period、flowering diameter、the number of flower/plant and the number of petal/flower which are associated with ornamental value it, such as in 13 main rugosa rose cultivars. The results showed that the differences between these ornamental traits were significant. The heritability (h2) of each trait was higher(0.5779-0.7462). Genetic variation coefficients were between 0.0315-0.5235. The variation patterns of flowering periods and their component units which are squaring period, initial blooming stage, florescence stage and faded stage considerable various. The combinations between flowering period and flowering diameter, the number of petal, or the number of flower were diverse. The cultivars‘Puce Dragon’,‘Tang Purple’,‘Tang Red’and‘Xizi’possessed larger number of flowers/plant and longer flowering period.‘Puce Dragon’,‘Purple Branch’,‘Tang Purple’and‘Xiangciguo’possessed larger number of petals and longer flowering period.‘Purple Branch’was the earliest and longest in flowering period, and multiple flowering. Then the CAR (n) model was developed. Finally, the breeding strategies of ornamental traits such as flowering period etc. in rugosa rose were discussed. This will provide theoretical basis for new cultivars breeding and parental mating design which will further improve the ornamental value of rugosa rose.7. The experiment studied the 2n pollen-induction and hybrid breeding in Rugosa rose. Based on the study on the meiotic process of pollen mother cells, it can be found that it is time when the petals were not exerted the bud (the petal is yellow, the anther is milk- white.) to induce 2n pollen. The hybrid result showed that there was considerable difference in fruit setting rate in different parent. This will provide reference for the selection of crossing parents in Rugosa rose breeding.更多还原