Report on variation in Brassica seed oil and protein content
Jackie Barker, Clare Hopkins, Sue Welham, Graham King (2008); Rothamsted Research; publication in preparation.
Background
Nuclear Magnetic Resonance (NMR) is widely used but only provides information on oil content, and is sensitive to moisture content (Krygsman & Barrett, 2004). As a result, we made use of non-destructive NIRS (Near Infrared Reflectance Spectroscopy; NIRSystem 5000, Foss) analysis on whole seeds, using facilities kindly provided by the plant breeding company KWS-UK (formerly CPB-Twyfords). NIRS is used routinely by the seed and breeding industry and has been the method of choice for genetic screening to date (Zhao et al., 2008). In addition to quantification of seed oil and protein, the spectra obtained can be analysed to provide information on other traits such as glucosinolates and sulphur content.Aims of this study
Modulation of the rapeseed oil/protein ratio can contribute to achieving a higher and more valuable oil yield for any given production input. In order to generate datasets that provide information on which to base decisions about the extent, distribution and location of genetic variability within the Brassica genepool for oil/protein ratio, and protein quality, lines (the majority of seed were kindly provided by WHRI) representing the BnaDFS, reference OSR mapping populations and other Brassica species were assessed. An important aspect of this task was to determine the feasibility of large-scale screening of genetic resources and genetic mapping populations. The task was to determine variation in a) overall seed composition, and b) protein composition.Materials and methods
Two types of data were collected: raw spectra and calibrated data (referenced to a particular OSR variety) for oil, protein, glucosinolates, sulphur and moisture content. By using specific calibration equations, the raw spectra can be dissected further to obtain both amino acid (Fontaine et al., 2001) and fatty acid (Sato, 2008) content information, though this was not within the remit or timescale of this project.Each sample was analysed in triplicate in randomised runs of 48 samples, with 2 controls per run to decrease machine/day variability. More than 1800 runs were performed. All calibrated data were represented as a linear mixed model analysed by REML (REsidual Maximum Likelihood), and estimates of heritability were obtained. The experimental design enabled us to identify and apportion sources of variation associated with line (heritability), accession, environment, sampling occasion, sample, batch and residual error.
In order to validate the NIRS results for oil content, and in cases where only small amounts of seed were available, we measured oil content using NMR instrumentation (MQC, Oxford Instruments) located at John Innes Centre, Norwich. The NMR required only ~200 mg seed for sampling, and so it was possible to select more uniform, high quality seed for analysis. More than 800 samples were analysed in triplicated randomised runs for the BnaDFS, and in a single run for the TVSL mapping population.
Oil content was determined using NMR instrumentation (MQC, Oxford Instruments) located at John Innes Centre, Norwich. The NMR required only ~200 mg seed for sampling, and so it was possible to select uniform, high quality seed for analysis. Smples were analysed in triplicated randomised runs for the BnaDFS.
Results
- Download Excel workbook of NMR data
- Download Excel workbook of NIRS data
- Within the BnaDFS, oil content measured by NIRS varied from 33.7-46.4% and protein from 19.2-28.8%.
- When assessed by NMR, oil content varied from 18.9-50.2%.
- The level of genotype x environment (G x E) interaction observed with the NIRS data indicates that care needs to be taken when assessing and ranking the range of diversity in material of different provenance.
