Skip to Main content Skip to Navigation
Theses

A Probabilistic Model of Flower Fertility and Factors Influencing Seed Production in Winter Oilseed rape (Brassica napus L.)

Abstract : The number of pods per plant and the number of seeds per pod are the most variable yield components in winter oilseed rape (WOSR). The production of a seed is the combination of several physiological processes, namely formation of ovules and pollen grains, fertilization of the ovules and development of young embryos, any problem in these processes may result in seed abortion or pod abortion. Both the number of ovules per pod and the potential for the ovule to develop into a mature seed may depend on pod position in the plant architecture and time of appearance. Furthermore, the expansion (basipetal) of ramifications is in inverse order of the initiation (acropetal) in WOSR. The complex developmental pattern of WOSR makes it difficult to analyze. In this thesis, we first investigated the variability of the following yield components (a) ovules/pod, (b) seeds/pod, and (c) pods/axis in relation to two explanatory variables. These two variables include (1) flower and inflorescence position and (2) time of pod appearance, linked to the effect of assimilate availability. Based on these experiments, we developed a probabilistic model to simulate the number of ovules per ovary and seeds per pod according to the biological phenomena of flower fertility. The model can predict the number of pollen grains per flower and distinguish the factors that influence the yield. Meanwhile, we compared our model to another model of flower fertility in kiwifruit developed by Lescourret et al., and improved the computation for the distribution of pollen grain number in our model.. In the last, we tested our model on the other species including cacao tree and soybean. The main contents are as follows: [1] The number and position of flowers and pods were recorded for the main stem (R0) and inflorescences R1, R4, R7, R9 and R11. The variety was Mendel. The results indicated that for the main stem, the number of ovules per pod decreased for a few ranks and then tended to increase and again to decrease at the end. On the ramifications R1 and R4, the number of ovules increased at first then remained constant with the pod rank, but it remained constant along the inflorescence on the other ramifications. However, the mean number of ovules per pod increased with ramifications from top to bottom. The number of seeds per pod did not vary with the pod rank at the basal positions and decreased afterwards along the inflorescence, but it did not differ between inflorescences. The number of ovules and seeds per pod did not vary with the time of pod appearance for the pods. However, the number of ovules and seeds per pod can be impacted by the time of pod appearance on the plant scale. [2] To analyze the effect of available assimilates on the yield components, different trophic states were created by clipping the main stem, ramifications or basal flowers. The results indicated that clipping the main stem or ramifications increased the number of ovules per pod, seeds per pod and pods per axis. However, clipping all the ramifications and basal flowers only increased the number of pods on the main stem. In addition, clipping treatments increased the mean seed weight. Accordingly, we can conclude that assimilate availability is one of factors influencing the seed production. [3] The model of flower fertility was developed by combining different probability distributions. The model can simulate the number of viable ovules per ovary, the number of pollen grains per flower, the probability of seed viability and the survival probability of pod. Model parameters were estimated using a Generalized Least Square method with two years, clipping treatments, pod ranks, ramification ranks and four varieties. The results indicated that ovule viability, the intensity of pollination, assimilate availability and architectural effects can influence the seed production.[4] Two resampling methods were used to analyze the stability of the model. The coefficients of variation (CV) using jackknifing were smaller than that using bootstrapping. However, the results of the two methods indicated that the parameters in the model were quite stable. The CVs of the parameters were small except the variance of the number of ovules per pod. The CV was a little large with 3.3 for jackknife and 11 for bootstrap, respectively. [5] The estimation for the distribution of pollen grain number was improved by comparing our model to the model of flower fertility in kiwifruit developed by Lescousrret et al. The model of flower fertiltiy in kiwifruit computed the number of fertilized ovules using stochastic method. The results were good but it was time-consuming. The model developed in the thesis assumed that the ratio of ovule and pollen was 1:1. The smaller value of them was taken as the number of fertilized ovules. However, the studies in other species found that one pollen grain might not be enough to fertilize one ovule. Thus, we introduced one parameter k to estimate the proportion of effective pollen grains in the model, in turns, to compute the distribution of pollen grain number. The results indicated that the model can simulate the flower fertility in WOSR very well. [6] The model of flower fertility was used to simulate the number of seeds per pod in soybean and cacao tree. The results in soybean were good. The number of ovules and seeds per pod can be well calibrated. For the cacao tree, the model can estimate the number of ovules and seeds per pod with good pollination. However, the estimations were not good for the situation of poor pollination. Based on the field experiments and model estimations, the following conclusions can be drawn. The amount of available assimilates was the primary determinant of pod and seed production during the flowering period. Furthermore, the ovule viability and pollination limitation could result in the decrease of the number of pods and the number of seeds per pod at the distal position of inflorescence. In addition, the distribution of resources was significantly affected by both the positions of pods within an inflorescence and the position of inflorescences within a plant in WOSR. The model of flower fertility could be a useful tool to study how to improve seed yield in flowering plants and the model can be applied to the other flowering plants
Document type :
Theses
Complete list of metadatas

Cited literature [147 references]  Display  Hide  Download

https://tel.archives-ouvertes.fr/tel-00612716
Contributor : Wang Xiujuan <>
Submitted on : Saturday, July 30, 2011 - 7:47:51 AM
Last modification on : Friday, April 10, 2020 - 2:10:27 AM
Long-term archiving on: : Monday, October 31, 2011 - 2:20:23 AM

Identifiers

  • HAL Id : tel-00612716, version 1

Collections

Citation

Wang Xiujuan. A Probabilistic Model of Flower Fertility and Factors Influencing Seed Production in Winter Oilseed rape (Brassica napus L.). Life Sciences [q-bio]. Ecole Centrale Paris; China Agricultural University, 2011. English. ⟨tel-00612716⟩

Share

Metrics

Record views

421

Files downloads

370