Phenotypic analysis

3.3.3.1 Near-Infrared Reflectance Spectroscopy

To determine seed quality traits about 3g of bulked seeds were scanned between 400 to 2498nm by a NIRS monochromator (FOSS NIR Systems model 6500, NIRSystems, Inc., Silversprings, MD, USA). Absorbance values log at 2nm intervals were recorded to create a NIR spectrum for each sample. With WinISI II software (version 1.50) seed quality traits were determined using the calibration raps2012.eqa provided by VDLUFA Qualitätssicherung NIRS GmbH (Teichstr. 35, 34130 Kassel, Germany, http://h1976726.stratoserver.net/cms/). Oil (Oil) and protein (Protein) content were determined in per cent at 91% seed dry matter.

Glucosinolate content (GSL) was measured in µmol/g at 91% seed dry matter. Fatty acid content (18:1, 18:3 and 22:1) was determined in per cent of the total fatty acid content.

3.3.3.2 Gas Chromatography

Fatty acid profile analysis was performed by gas chromatography according to the method of Thies (1971) as described by Rücker and Röbbelen (1996) with modifications as follows: Two 4mm stainless steel metal balls and 200mg seeds per genotype were filled into a 3ml polypropylene tube. 1000µl of Na-methylate in methanol (0.5mol/l) were added to each tube and tubes were tightly closed with a screw cap. To extract and transmethylate fatty acids seeds were milled by 2min shaking using a custom-built vertical homogenizer (Institute of Applied Plant Nutrition, Georg-August-Universität Göttingen, Göttingen, Germany). Grounded seeds were then incubated for 15min at room temperature while shaking every 5min. 300µl 5% NaHSO4 as well as 500µl isooctane were added and tubes were vortexed for salt precipitation and extraction of fatty acid methyl ester in isooctane followed by 1min centrifugation at 1000rpm (150 x g). 200µl of the upper phase were pipetted into septum vials and closed with crimp caps.

2µl of the extract were injected into a gas chromatograph (Trace GC ultra, Thermo Electron corporation) with a 25m x 0.25mm I.D. FFAP column (Macherey & Nagel, 0.25µm film thickness, 210°C, split injection 1:70, carrier gas: 150kPa H2, injection/detector: temperature 230°C).

Palmitic (16:0), palmitoleic (16:1), hexadecadienoic (16:2), stearic (18:0), oleic (18:1), linoleic (18:2), linolenic (18:3), arachidic (20:0), eicosenoic (20:1), behenic (22:0) and erucic (22:1) acid were determined by the GC chromatogram of each sample. Fatty acids were expressed as per cent of the sum of all fatty acids.

3.3.3.3 Adjustment of NIRS predicted erucic acid contents

Comparing NIRS predictions and GC measures of erucic acid content within the SGEDH population a conspicuous underestimation of erucic acids contents of NIRS data was identified (cf. Figure 3.1a). Comparing the differences between GC and NIRS values, different deviations were observed within the group of genotypes with no or low erucic acid, determined as erucic acid free, and the one of genotypes with medium and high erucic acid. Because of the different deviations both groups were adjusted individually. Erucic acid free genotypes, showing generally negative NIRS predictions, were adjusted by subtracting a group specific correction factor of -6.35%. This factor was calculated as the mean difference between GC and NIRS erucic acid content of means over 14 locations of 70 genotypes showing no or low erucic acid content determined by GC. NIRS erucic acid content of genotypes with medium and high GC erucic acid content were adjusted using the regression equation of this group, y = 1.50x – 5.05 (Figure 3.1b).

The comparison of adjusted NIRS predicted erucic acid contents and GC erucic acid contents (Figure 3.1c) showed a high coefficient of determination with R² = 0.99.

3.3.3.4 Thousand kernel weight

Thousand kernel weight (TKW) was obtained from weight conversion of 500 seed. Seeds were counted using a Contador seed counter (Pfeuffer GmbH, D-97318 Kitzingen, http://www.pfeuffer.com).

3.3.3.5 Protein content and glucosinolate content in the defatted meal

Protein content in the defatted meal (Prot.idM) was calculated by using NIRS predicted seed oil and seed protein content (both at 91% dry matter) as:

% protein in the defatted meal (Prot.idM) = [% protein / (100 - % oil)] * 100

Glucosinolate content in the defatted meal (GSLidM) was calculated by using NIRS predicted seed oil and seed glucosinolate content (both at 91% dry matter) as:

µmol/g glucosinolates in the defatted meal (GSLidM) = [µmol/g glucosinolates / (100 - % oil)] * 100

3.3.3.6 Correction of oil content considering erucic acid content

SGEDH population segregated for erucic acid content, furthermore erucic acid and oil content showed a strong positive correlation. Thus, to be able to compare the oil content of genotypes with varying erucic acid contents, oil contents were corrected by eliminating the effect of erucic acid on oil content in three different ways.

Figure 3.1: xy plots of (a) erucic acid content measured by gas chromatography and NIRS predicted erucic acid content for whole SGEDH population, (b) for genotypes with medium and high erucic acid contents, and (c) erucic acid content measured by gas chromatography and adjusted NIRS values of the SGEDH population. Data are presented as means over 14 environments

y = 1.11x + 5.51

adjusted NIRS 22:1 content in % (91% DM) (c)

Correction by regression

For the correction of oil content by regression information, the linear regression between GC erucic acid content and NIRS oil content from trait means over 14 locations was calculated. The slope of this regression was used to calculate the regression corrected oil content the following way:

NIRS oil content – (slope of linear regression between GC erucic acid content and NIRS oil content * GC erucic acid content)

Correction by molecular weight

De novo fatty acid biosynthesis until CoA takes place in the chloroplasts. However, oleoyl-CoA elongation to erucoyl-oleoyl-CoA via eicosenoyl-oleoyl-CoA and triacylglycerol (TAG) assembly takes place in the cytoplasm. If one molecule oleoyl-CoA is elongated to eicosenoyl-CoA and subsequently to erucoyl-CoA, this leads to an increase of molecular weight of 10 and 19%, respectively. Therefore, if the number of fatty acid molecules produced by the de novo fatty acid biosynthesis remains the same an increase in erucic acid content in the seed oil should results in a proportional increase in seed oil content. Following this simplified theoretical assumption, the molecular weight of eicosenoic acid and erucic acid can be reduced on the basis of oleic acid by 9 and 16.6% to eliminate their effects on oil content. Hence the molecular correction of oil content was calculated from trait means over 14 locations as:

NIRS oil content – [0.09 * (GC eicosenoic acid content * NIRS oil content / 100) + 0.166 * (GC erucic acid content * NIRS oil content / 100)]

Correction by conditioning

The mixed model approach for the conditional analysis of quantitative traits described by Zhu (1995) was applied to calculate oil content independent of erucic acid contents. The conditioning allows analysis of correlated traits independently of variation in the secondary trait. This method is very similar to the estimation of adjusted values in a covariance analysis.

Required software tools were provided by Prof. Jianyi Zhao (Zhejiang Academy of Agricultural Sciences, Hangzhou, PR China). The software was used to calculate the conditional phenotypic

values (cond) from trait means over 14 locations of NIRS predicted oil content at 91% seed dry matter conditioned by erucic acid content determined by gas chromatography.

3.3.3.7 Correction of protein content in defatted meal considering erucic acid content

The SGEDH population showed a strong positive correlation between erucic acid and protein content in defatted meal. To investigate protein content in defatted meal independent of the effect caused by erucic acid, regression and conditional correction of protein content in defatted meal was conducted (cf. section 3.3.3.6).

Correction by regression

For the correction of protein content in defatted meal by regression information, the linear regression between GC erucic acid content and protein content in defatted meal was calculated.

The slope of this regression was used to calculate the regression corrected protein content in defatted meal the following way:

Protein content in defatted meal – (slope of linear regression between GC erucic acid content and protein content in defatted meal * GC erucic acid content)

Correction by conditioning

Conditional correction of protein content in defatted meal was conducted using the method of Zhu (1995) as explained in section 3.3.3.6.