Fatty acids composition of oilseed rape genotypes as affected by solar radiation and temperature

Field Crops Research 212 (2017) 165–174

Contents lists available at ScienceDirect

Field Crops Research journal homepage: www.elsevier.com/locate/fcr

Fatty acids composition of oilseed rape genotypes as affected by solar radiation and temperature

MARK

Marion Gauthiera, Didier Pelleta, Corinne Monneya, Juan Manuel Herreraa, Marie Rougiera,b, ⁎ Alice Bauxa, a b

Department Field Crops, Agroscope, Changins, 1260 Nyon, Switzerland École nationale supérieure agronomique de Toulouse, 31 326 Auzeville Tolosane, France

A R T I C L E I N F O

A B S T R A C T

Keywords: Oilseed rape Fatty acid composition High oleic low linolenic Temperature Solar radiation

Environmental parameters are known to affect the fatty acid profiles of oilseed crops. Whereas the role of temperature in fatty acid desaturation is well documented, the impact of solar radiation, and therefore photosynthetic activity, is less known. Moreover, the interaction between temperature and solar radiation has rarely been documented. This study is based on field data and an experiment under semi-controlled conditions. The first experiment in growth chambers was designed to assess independently the impact of light intensity and temperature; the analysis of the data from field experiments with natural variation in solar radiation and temperature showed their relation to the alpha-linolenic acid concentration (C18:3) at different growth stages. Temperature affected the fatty acid composition, by increasing the oleic acid concentration (C18:1) and decreasing the C18:3. The results validated with more years and genotypes that the period from 680 to 930°-days after onset of flowering (DDAF) was sensitive to the minimum temperatures. Solar radiation during seed filling was also found to be negatively correlated with C18:3. However, during an earlier sensitive period, from 100 to 300 DDAF, solar radiation had a positive impact on C18:3 concentration. These seemingly contradictory effects could be the consequence of the impact of solar radiation on the source-sink ratio, with a higher source-sink ratio leading to lower C18:3 concentration. Moreover, our data showed that the response of the C18:3 concentration to solar radiation and temperature depended on the genotype with a tendency to less intense effects on high-oleic low-linolenic (HOLL) genotypes. As a result of these findings, an empirical model with solar radiation from 100 to 300 DDAF and the minimum temperature from 680 to 930 DDAF, with genotype-dependent sensitivity, was proposed to predict C18:3.

1. Introduction The use of oils depends on their fatty acid composition. Conventional oilseed rape (OSR) has about 60% C18:1 and 10% C18:3. Its nutritional quality is excellent, due to its high polyunsaturated fatty acids concentration, and low saturated fatty acids concentration, but the same reasons make it unstable at high temperatures (Mollers, 2002). In contrast, high-oleic low-linolenic (HOLL) OSR varieties are characterized by high oleic acid concentration (C18:1 > 75%) and low linolenic acid concentration (C18:3 < 3.5%), and therefore a better oxidative stability at high temperature (Carré et al., 2007). HOLL oilseed rape can therefore be used for deep frying without hydrogenation (Matthäus, 2007). This post-harvest treatment tends to produce trans fatty acids which raise concerns from consumers about the health

effects (Ascherio and Willett, 1997; Koletzko and Decsi, 1997). It has been restricted in Switzerland since 2008 (Federal Office of Public Health, 2008). Consequently, the Swiss oil industry had a special interest on HOLL OSR oil and insisted on the very low C18:3 concentration, in order to secure oxidative stability. Fatty acid composition is the balance between saturated, monounsaturated and polyunsaturated fatty acids, composing the triacylglycerol, storage form in oil crops. Photo-assimilates are imported into the plastids where fatty acids are elongated by the fatty acid synthase enzyme complex. Most of the stearic acid (C18:0) is then desaturated into C18:1 which is in a large part exported to the cytosol. A fraction of C18:1 is then desaturated into C18:2, and C18:2 into C18:3 through the action of the fatty acid desaturases (FAD) 2 and 3 respectively (Singer et al., 2016). A minor part of C18:1 is also desaturated in

Abbreviations: OSR, oilseed rape; HOLL, high-oleic low-linolenic; C18:3, alpha-linolenic acid concentration; C18:2, linoleic acid concentration; C18:1, oleic acid concentration; Tmin, minimum temperature; Rad, solar radiation; PTQ, photothermal quotient; DDAF, degree-days after the onset of flowering; FAD, fatty acid desaturase ⁎ Corresponding author. E-mail address: [email protected] (A. Baux). http://dx.doi.org/10.1016/j.fcr.2017.07.013 Received 13 January 2017; Received in revised form 6 July 2017; Accepted 20 July 2017 Available online 02 August 2017 0378-4290/ © 2017 Elsevier B.V. All rights reserved.

Field Crops Research 212 (2017) 165–174

M. Gauthier et al.

mid-flowering (Leterme, 1985), which could results in difference among species. The goals of this study are to determine whether solar radiation affects C18:3 concentration in OSR, to assess the role of the genotype in the response of C18:3 concentration to solar radiation and temperature, and to determine whether solar radiation can be used to improve the prediction of C18:3 based on the temperature.

the plastids into C18:2 with FAD6, and then C18:2 can be desaturated by either FAD7 or FAD8 into C18:3 (Ohlrogge and Browse, 1995). Mollers (2002) reported that the activity of the desaturase enzymes mediating this reaction is affected during the breeding of HOLL OSR lines, in particular with a modification of the amino acid sequence of FAD2. In HOLL cultivars, the lower concentration of C18:2 and C18:3, compared to conventional genotypes, is due to lower activity of both oleate and linoleate desaturases (Baux et al., 2013). Beside the genotypic variability, environmental factors such as temperature, solar radiation, water deficit, salinity, oxygen and soil nutriments, also influence the oil composition of oilseed crops (Singer et al., 2016). The impact of temperature on C18:3 concentration in OSR has been demonstrated by many studies (Canvin, 1965; Trémolières et al., 1982; Champolivier and Merrien, 1996; Deng and Scarth, 1998; Pritchard et al., 2000; Baux et al., 2008; Baux et al., 2013). In addition, a sensitive period has been identified, as well as interactions between temperature and genotype, for several species. Baux et al. (2013) showed that minimum temperatures from 680 to 930°-days (base 0 °C) after the onset of flowering (DDAF) had a negative impact on C18:3, and this effect was less intense for HOLL genotypes than for conventional genotypes, partly because the desaturation of C18:2 into C18:3 was no longer temperature dependent in mutated genotypes. The effect of the night temperature on C18:1 has also been reported for other oilseed crops, such as sunflower (Izquierdo et al., 2002), maize, and soybean (Zuil et al., 2012). For the sunflower, Izquierdo and Aguirrezábal (2008) found that the relation between the minimum night temperature and C18:1 was genotype-dependent. The impact of the temperature was explained by thermal regulation of the enzymes involved in the fatty acid biosynthesis (Wilmer et al., 1998; Garcia-Diaz et al., 2002). Intercepted solar radiation also affected the fatty acid profile in OSR, soybean, maize, and sunflower, when modified by shading and thinning (Izquierdo et al., 2009). These authors showed that during grain filling, it impacted C18:1 positively and C18:3 negatively in OSR. However, Wang et al. (2016) found that C18:3 decreased when the pods were shaded during seed development. The interaction between solar radiation and temperature was also studied for other species. In controlled conditions, Echarte et al. (2010) found that the night temperature and the intercepted solar radiation had an additive effect on C18:1 in sunflower. C18:1 was sensitive to these environmental factors during distinct critical periods (Echarte et al., 2013). The intercepted solar radiation and the minimum temperature during grain filling were used to predict the C18:1 of the high oleic genotypes of maize and soybean (Zuil et al., 2012). Variations in solar radiation and temperature affect the source-sink balance and explain the differences in seed yield (Diepenbrock, 2000; Jullien et al., 2011; Weymann et al., 2015) and oil content (Canvin, 1965; Willms et al., 1999). Impact on oil composition have also been studied on sunflower. Echarte et al. (2012) proposed a conceptual model that explains changes in the fatty acid composition by the impact of the intercepted solar radiation on assimilate availability. It was suggested that for high assimilate levels, the oleic acid desaturase was saturated, leading to an accumulation of C18:1. Conversely, at low assimilate levels, the enzymatic activity would no longer be the limiting factor, and thus, the percentage of C18:2 would increase. The intercepted solar radiation would regulate this equilibrium by determining the assimilate availability. In addition, Durruty et al. (2016) built a kinetic model for the sunflower to simulate grain filling and fatty acid biosynthesis. This model describes the enzymatic phases of the steps leading to the biosynthesis of oleic acid and its desaturation into linoleic acid. The influence of environmental factors on fatty acid composition is better characterized for the sunflower than for OSR, and the potential impact of genotypic variability in OSR is unknown. Moreover, in OSR, the source of photo-assimilates changes between pod formation and seed filling, as the pods become autotrophic after about 300 ° days after

2. Materials and methods This study consists of two distinct parts with complementary objectives. First, because weather variables are strongly correlated (Bristow and Campbell, 1984; Prieto et al., 2009), an experiment in semi-controlled conditions was set up to test independently the effect of solar radiation and temperature on fatty acid composition. Second, a large dataset with multiple years and sites exposed to natural variations in solar radiation and temperature was analyzed to investigate the impact of the minimum temperature and solar radiation on fatty acid composition at the actual field scale. 2.1. Experiment in semi-controlled conditions This experiment took place in Changins, Switzerland, and was conducted twice during the 2012/13 and 2013/14 crop seasons. A factorial design was used to evaluate independently the impact of the temperature and photosynthetically active radiation (PAR) on the fatty acid composition of three winter OSR varieties, two HOLL varieties (V141OL, V316OL), and one very low linolenic variety (without the high oleic mutation: MSP21), bred by Monsanto (St. Louis, USA) and DSV (Lippstadt, Germany). The plants were sown on September 5, 2012, and September 13, 2013, in a mix of 18% sterile topsoil, 24% coco peat, 26% blond peat, and 32% black peat in 5 L pots. The seeds were sown to reach a density of three seedlings per pot. During the growth, nitrogen was supplied in a quantity corresponding to 120 kg/ha in the spring (1.12 g MgSAmmonsalpeter 25% (Lonza, Basel, Switzerland) and 1 g ammonium nitrate 27% (Landor, Muttenz, Switzerland) per pot). Fungicide was applied in March 2013 (Horizon EW®, Bayer, 1 L/ha) at the phenological stage 32 of the BBCH scale (Lancashire et al., 1991). Insecticides was used against pollen beetles, Meligethes aeneus, in April 2013 (BBCH stage 59, Karatezeon®, Syngenta, 75 mL/ha) and in March 2014 (BBCH stage 55, Blocker®, Oyma AGRO, 0.3 L/ha) and against aphids (Myzus persicae; Pirimor®, Syngenta, 0.5 kg/ha) three times in May and June 2013 and 2014 (at the end of the flowering period). From sowing until the beginning of the flowering period in April, the pots were kept outside but were protected from frost with a cold frame since the beginning of December. Inside the frame, the temperature was regulated with an isolating cloth and a small heater when needed to avoid frost damage. From flowering to maturity in July, the pots were moved into growth chambers and submitted to four treatments varying in temperature and solar radiation (Table 1). Two phytotrons (CMP3244, Conviron, Isleham, UK) were used, at a plant density of 48 plants per square meter, which was consistent with local field conditions. They were illuminated with 160W cool white fluorescent lamps (model F72T12-CW-VHO, Osram Sylvania, Mississauga, Canada). The photoperiod was fixed at 14 h inside the phytotrons. Each level of the temperature treatment was allocated to one phytotron. To apply two radiation levels, each phytotron was split into two parts of equal size with an opaque pane. In half of the phytotrons, light was partly shaded using aluminum strips. PAR in the 400–700 nm waveband was measured at the top of the canopy in μmol m−2 s−1 with a ceptometer (AccuPAR LP-80, Decagon Devices, Pullman, USA). Measured PAR was converted into W m−2 (from photon to energy units) using a factor of 0.22, which is suitable for this light source (Barta et al., 1992). In addition to the build-it sensor, the temperature was recorded every hour in each half-phytotron with a data logger (LogTag, Oceasoft, 166

Field Crops Research 212 (2017) 165–174

M. Gauthier et al.

composition of the oil with gas chromatography (Thermo Fisher Scientific Inc., Waltham, USA) and of the oil concentration (expressed at 6% humidity) with near-infrared spectrometry (Foss NIR System Inc., Silver Spring, USA).

Table 1 Environmental conditions measured in the growth chamber ± standard deviation. Treatment

Day temperature (°C)

Night temperature (°C)

PAR (W m−2)

D23–N15 full radiation D23–N15 shading D18–N10 full radiation D18–N10 shading

23.7 ± 4.32

14.6 ± 0.38

83 ± 18

24.1 ± 1.05 20.5 ± 1.05

15.3 ± 0.15 10.3 ± 0.07

38 ± 4.0 84 ± 22

19.0 ± 0.15

10.4 ± 0.17

32 ± 9.0

2.2.2. Dataset The data were selected according to the following rules: i) Only isolated data were used for HOLL genotypes, ii) the genotypes that had fewer than eight observations were excluded, and iii) sites with fewer than 15 observations and/or for which daily weather data were not available from nearby (< 10 km) meteorological stations were excluded. The final dataset was unbalanced (Table 2) and included 261 observations (58 for the HOLL genotypes, 210 for the conventional genotypes) over 11 years, 7 locations, 40 environments (year × location), and 18 genotypes (including four HOLL genotypes). Daily weather data were obtained from the closest weather stations (MétéoSuisse) taking into account the elevation of the experimental sites. We calculated the daily photothermal quotient (PTQ), metric of the availability of photosynthates (Leterme, 1985), as the ratio between the amount of solar radiation and the mean temperature above 0 °C. To determine during which stage of grain filling C18:3 was the most sensitive to Tmin and Rad, these weather variables were averaged over several time windows, defined by degree-days (base 0 °C) from flowering onward. The use of degree-days was chosen because they are a good proxy for the phenological stages (Habekotté, 1997). All the time windows with a length of 100, 200, and 300°-days were computed from 0 to 1000 DDAF. For both variables, the average value from 680 to 930 DDAF was also computed as this window was identified in a previous study (Baux et al., 2013) as a high-sensitivity period of C18:3 to Tmin. Thirty-one time windows were tested.

D23–N15: 23 °C during the day and 15 °C during the night; D18–N10: 18 °C during the day and 10 °C during the night. PAR: Photosynthetically active radiation measured during the light period (photoperiod 14 h).

Montpellier, France) to control for micro-environmental differences and potential failures. Plants were manually harvested at physiological maturity (BBCH stage 97) which spanned one month due to the genotype diversity and the environmental treatments. Seeds from the same pot were grouped and oven-dried (24 h at 60 °C). The fatty acid composition of the oil was measured with gas chromatography (Thermo Fisher Scientific Inc., Waltham, USA) and the oil concentration (expressed at 6% humidity) with near-infrared spectrometry (Foss NIR System Inc., Silver Spring, USA). For each treatment (temperature × radiation combination), there were four pots per genotype. As within each treatment the four pots were not independent repetitions, we averaged the pots’ fatty acid composition and oil concentration to avoid pseudo-replication that would artificially inflate the statistical power of the study. 2.2. Field data

2.3. Data analysis and modeling 2.2.1. Trials The field data were obtained from several experiments set up for evaluation of the OSR varieties between 2005 and 2015, testing about 25 commercialized and candidate varieties in seven locations in Switzerland (Table 2). Because not all genotypes performed well enough to remain in the trials, the genotypes were not all tested at each location and in each year (Table 2). Since 2010, HOLL varieties were separated from conventional ones to prevent cross pollination (Baux et al., 2011). The plot size was 29.25 m2 (2.25 × 13 m), and there were three replications. Crop management (herbicide treatment, pest control, and N fertilization, plant density) followed common practices for OSR in Switzerland. No fungicide was sprayed. The flowering date, corresponding to the phenological stage 61 of the BBCH scale, was scored for each genotype. At full maturity, the plots were mechanically harvested. Then, the seed yield and the thousand kernel weight were measured after oven-drying (24–48 h, until constant weight was reached, at 60 °C). A composite sample of the seeds was assembled for each triplet (genotype, year, location) for the analysis of the fatty acid

Data were analyzed using the statistical software R (R Core Team, 2015). Variations in fatty acid composition (i.e., C18:1, C18:2, and C18:3 concentrations) and oil content in the experiments under semicontrolled conditions were studied using a mixed model analysis of variance with the year as a random effect. The genotype, the temperature, the amount of solar radiation, and their interactions were set as fixed factors. Interactions that were not statistically significant (p > 0.05) were excluded one by one, starting from the highest pvalue. Then, differences between treatment levels were assessed with a pairwise comparison of the least squares means (Lenth, 2016). For the field data analysis, the R package lme4 (Bates et al., 2015) was used to perform a linear mixed effects analysis of the relation between C18:3 and the weather variables. We included the year-site combination as random effects in order to take into consideration that observations might come from the same environment. We used the least squares means (Lenth, 2016) to perform a comparison of the factors’ levels. The correlation coefficients reported throughout were obtained

Table 2 Description of the trial locations in the final dataset from the field experiments. Location

Elevation (m)

Latitude

Longitude

Range of sowing dates

Range of flowering dates

Length of the cropping season (days)

Tmin_680_930 (°C)

Burtigny Changins Gennersbrunn Goumoens Missy Oensingen Reckenholz

720 425 440 618 442 485 440

46.47° 46.40° 47.71° 46.65° 46.88° 47.29° 47.43°

6.26° 6.26° 8.69° 6.60° 6.97° 7.71° 8.52°

Aug 26–Sep 01 Aug 24–Sep 05 Sep 01–Sep 03 Aug 23–Sep 03 Aug 29 –Sep 03 Aug 29–Sep 06 Aug 26–Sep 11

Apr Apr Apr Apr Apr Apr Apr

324 309 319 322 321 315 313

11.1 12.4 13.8 14.1 12.3 11.7 12.2

N N N N N N N

W W W W W W W

8–Apr 26 5–Apr 27 17–May 2 8–May 4 8–May 3 5–Apr 27 4–May 2

± ± ± ± ± ± ±

6.40 8.50 2.90 5.50 9.20 6.60 8.50

± ± ± ± ± ± ±

1.10† 1.68 2.14 2.15 1.10 1.21 1.95

† Tmin_680_930: daily minimum temperature averaged during the period from 680 to 930°-days after flowering ± standard deviation. Values between brackets are for the high oleic low linolenic (HOLL) genotypes.

167

Number of years

Total number of genotypes

4 8 3 8 4 5 8

12 (4)† 18 (4) 8 (0) 18 (4) 9 (4) 17 (4) 14 (0)

Field Crops Research 212 (2017) 165–174

M. Gauthier et al.

Table 3 General structure of the tested mixed models. The models differed in the structures determined by the weather variable considered and by the time windows from which the weather data were obtained. Weather Variables

Models’ equations

Number of tested models

Minimum temperature Solar radiation Photothermal quotient Minimum temperature and solar radiation

C18:3i = μ + Tmini + Gi + Tmini × Gi + (Site × Year)i + ϵ i C18:3i = μ + Radi + Gi + Radi × Gi + (Site × Year)i + ϵ i C18:3i = μ + PTQi + Gi + PTQi × Gi + (Site × Year)i + ϵ i C18:3i = μ + Tmini + Radi + Gi + Tmini × Gi + Radi × Gi + (Site × Year)i + ϵ i

31 31 31 961

C18:3i: linolenic acid concentration associated with one triplet (Site × Year × Genotype)i; μ: mean population of C18:3; Tmini: daily minimum temperature (°C) associated with the couple (Site × Year)i average over a defined time window; Radi: daily solar radiation (W m−2) associated with the couple (Site × Year)i average over a defined time window; PTQ (photothermal quotient)i: daily photothermal quotient (W m−2 °C−1) associated with the couple (Site × Year)i average over a defined time window; Gi: genotype of the observation i; (Site × Year)i is considered as random, all the other effects are fixed; ϵ i: residual error.

with the Pearson correlation as implemented in R. Each weather variable was measured for different time windows and tested according to the general equations in Table 3. When Tmin and Rad were combined, the best period in terms of accuracy to predict C18:3 might be different than taking each variable independently. Therefore, we also tested each combination of periods. Before we analyzed the results for the models, a visual inspection of the residual plots did not reveal any obvious deviations from homoscedasticity or normality. The goodness of fit of the models was measured using the root mean square error (RMSE) and an R2-like index defined as the R2 of the linear regression of the fitted values and the observations of C18:3 (Byrnes, 2008). These indices were computed using fits at the population level, i.e. for any Site × Year combination setting the random effect to zero. Therefore, the indices assess the ability of the fixed effects to explain the variability in C18:3. The prediction accuracy of the most promising models was assessed using the root mean square error of prediction (RMSEP), estimated from a cross-validation procedure taking into account the dependences of the sites and the years (Landschoot et al., 2012). In order to help to understand the mechanism involved, other parameters like C18:1, C18:2, oil concentration, seed yield, thousandkernel weight, and number of seeds per square meter were also investigated using the same model structure.

was higher for the high-radiation treatment with, on average, 43% of oil vs. 37% in the shading treatment (Fig. 1), and, for the higher-radiation treatment only, the high-temperature D23–N15 treatment was associated with a significantly lower oil concentration (41% vs. 44% for the D18–N10 treatment). C18:3 concentration was significantly affected by the genotype (p < 0.0001), the temperature (p < 0.001), the amount of radiation (p < 0.001) and the interaction between radiation and genotype (p < 0.01). On average, the low-temperature D18–N10 treatment resulted in approximately 0.6 percentage points more C18:3 than the D23–N15 treatment (p < 0.001). The genotype MSP21 had lower C18:3 than the others, and was the only one responding to the lowradiation treatment with a significant increase in C18:3 concentration (35% increase, p < 0.01, Fig. 1). The radiation treatment did not have significant impact on C18:1 or C18:2 (p > 0.05). C18:2 was only affected by the genotype (p < 0.001), with higher concentration for MSP21 (23% vs. 11% concentration for V141OL and MDS16), while C18:1 concentration was affected by the genotype (p < 0.001) and the temperature (p < 0.05). The low-temperature D18–N10 treatment was associated with lower C18:1 (74% vs. 77% for the D23–N15 treatment). C18:1 was also lower for MSP21 than for other genotypes (67% vs. 80% for V141OL and V316OL).

3. Results

3.2. Relation between C18:3 and environmental conditions in the field

3.1. Effects of temperature and radiation on oil composition under semicontrolled conditions

The field data allowed us to study the effect on C18:3 of the temperature and the amount of solar radiation within the range of these variables that occur on farms throughout Switzerland (Table 2). The average seed yield was 3.38 t/ha (0.84 standard deviation), and the average oil concentration was 44.6% (3.35 percentage points as the standard deviation) at 6% humidity.

The analysis of the variance did not reveal any effect of the genotype on the oil concentration (p > 0.05), but statistically significant deviations in the oil concentration were associated with the interaction between the temperature and the amount of radiation (p < 0.05). It

Fig. 1. Oil and linolenic acid concentrations per radiation x temperature treatment and genotype according to the analysis of variance. D23–N15: 23 °C during the day and 15 °C during the night; D18–N10: 18 °C during the day and 10 °C during the night Errors bars represent the confidence interval (95%).

168

Tmin_680_930

Tmin_680_930*

Tmin

Rad_100_300

Rad_100_300 Rad_700_1000

Rad

PTQ_400_600

PTQ

x

x

Tmin × G

X X

Rad × G

Variables in the C18:3 model added to the general structure: C18:3i = μ + … + (Site × Year)i + ϵ i PTQ × G 0.9161 0.9655 0.9476 0.9329 0.9318 0.9703

R -like

2

0.7456 0.4721 0.5830 0.6648 0.6726 0.4376

RMSE

Whole dataset

0.8516 0.6352 0.7229 1.1121 0.8062 0.5861

RMSEP 0.4588 0.6380 0.5986 0.4607 0.5366 0.6550

R -like

2

0.4462 0.2954 0.3459 0.4265 0.4301 0.2789

RMSE

HOLL genotypes

0.4590 0.4024 0.4579 0.4677 0.4542 0.4290

RMSEP 0.2671 0.7111 0.5677 0.4337 0.4192 0.7523

R2-like

0.9322 0.6863 0.7811 1.2333 0.8797 0.6229

RMSEP

169

V280OL MDS10 Visby

V141OL

DSV Monsanto NPZ

Monsanto

KWS

Dekalb

Breeder

24 10 27

13

16

12

Number of observations

3.50 ± 0.66 4.10 ± 1.10 10.4 ± 0.71

4.10 ± 0.90

11.3 ± 0.77

12.9 ± 0.66

Intercept (%)

a a bc

a

bc

c†

*** *** ***

***

***

***††

a ab ab b ab b

−0.34 ± 0.048 −0.23 ± 0.063 −0.08 ± 0.076 −0.10 ± 0.054 −0.12 ± 0.094 −0.10 ± 0.058

Slope (% °C−1)

C18:3i = μ + Tmin_680_930i + Gi + Tmin_680_930i × Gi + (Site × Year)i + ϵ i

*** ** ***

**

***

***

1.7 ± 0.51 2.4 ± 0.64 8.0 ± 0.46

2.6 ± 0.62

6.3 ± 0.66

5.3 ± 0.68

Intercept (%)

a a bc

a

bc

c

*** *** ***

***

***

***

0.003 ± 0.0023 0.001 ± 0.0028 0.005 ± 0.0021

0.002 ± 0.0027

0.010 ± 0.0030

0.014 ± 0.0030

Slope (% W−1 m−2)

ab a abc

ab

abc

c

C18:3i = μ + Rad_100_300i + Gi + Rad_100_300i × Gi + (Site × Year)i + ϵ i

*** NS ***

*

***

***

††

Letters indicate statistically significantly different groups of coefficients from the pairwise least square means comparison with the Tukey method. Test whether the coefficient is statistically significantly different from zero: ***: p < 0.001; **: p < 0.01; *: p < 0.05; NS: not statistically significant. C18:3i: linolenic acid concentration i associated with one triplet (Site × Year × Genotype); μ: mean population of C18:3; Tmin_680_930i: daily minimum temperature (°C) associated with the couple (Site × Year)i averaged over the period from 680 to 930°-days after the onset of flowering; Rad_100_300i: daily solar radiation (W m−2) associated with the couple (Site × Year)i averaged over the period from 100 to 300°-days after the onset of flowering; Gi: genotype of the observation i; (Site × Year)i is considered random, and all the other effects are fixed; ϵ i: residual error.

†

Conventional hybrid Conventional hybrid HOLL Open Pollinated HOLL hybrid HOLL hybrid Conventional hybrid

Cormoran

Hybrirock

Type

Genotype

Table 5 Effect of the minimum temperature averaged from 680 to 930°-days after flowering (DDAF) on the linolenic acid concentration (mod0) and the effect of solar radiation averaged from 100 to 300 DDAF on the linolenic acid concentration (mod1) for five oilseed rape genotypes (18 in total in the dataset).

0.8097 0.5107 0.6336 0.7173 0.7260 0.4725

RMSE

Conventional genotypes

Tmin: daily minimum temperature (°C). Rad: daily solar radiation (W m−2). PTQ: Photothermal quotient (W m−2 °C−1). Tmin_680_930 is the minimum temperature averaged over the period from 680 to 930°-days after the onset of flowering. Additional weather variables averaged over specific periods are noted as following the rule: NameOfVariable_StartWindow_EndWindow. G: genotype. C18:3i: linolenic acid concentration i associated with one triplet Site × Year × Genotype; μ: mean population of C18:3; (Site × Year)i is considered random, and all the other effects are fixed; ϵi: residual error. x: indicates that the interaction of the variable (Tmin, Rad or PTQ) and the genotype is statistically significant. *:daily minimum temperature averaged from 680 to 930°-days after flowering.

Modnull mod0 mod1 mod2 mod3 mod4

Model

Table 4 Summary of the linolenic acid concentration (C18:3) mixed models investigated. Several variables averaged over different time windows were tested as fixed effects in addition to the general structure with the environment as random effect.

M. Gauthier et al.

Field Crops Research 212 (2017) 165–174

Field Crops Research 212 (2017) 165–174

M. Gauthier et al.

the RMSEP from 0.85 to 0.72. It improved mainly the prediction accuracy for the conventional genotypes (Table 4). No statistically significant relation was found between Tmin and C18:3 for the 100–300 DDAF period and no correlation between Tmin and Rad (r = 0.02 and p > 0.1).

3.2.1. Relation between C18:3 and minimum temperature C18:3 was more sensitive to Tmin at late periods. In particular, the period from 680 to 930 DDAF was the one for which the average minimum temperature better explained variations in C18:3 (mod0 in Table 4, R2-like index = 0.966). The likelihood ratio test between a model without Tmin (modnull in Table 4) and mod0 indicated that Tmin_680_930 had a statistically significant effect on C18:3 (p < 0.001). Moreover, the test also confirmed that the effect of Tmin_680_930 on C18:3 depended on the genotype (the likelihood ratio test, p < 0.01). Overall, increases in Tmin between 680 and 930 DDAF had a negative impact on C18:3. The slopes varied from −0.34 ( ± 0.048) for Cormoran to −0.08 ( ± 0.076) for V141 OL (Table 5). The analysis revealed two groups with statistically significantly different slopes; the HOLL genotypes had coefficients closer to zero, compared to the conventional genotypes. However, no clear distinction was found between the temperature sensitivity of the HOLL genotypes and the conventional genotypes because the conventional cultivar Visby was also barely sensitive to Tmin_680_930. Adding Tmin_680_930 to modnull (with the genotype as the only fixed effect) decreased the RMSEP from 0.85 to 0.63 (Table 4)

3.2.2.2. Solar radiation from 700 to 1000 DDAF (Rad_700_1000). Rad_700_1000 had a statistically significant overall effect on C18:3 (p < 0.05) that depends on the genotype (p < 0.05). The effect of Rad_700_1000 on C18:3 was not statistically significant for the HOLL genotypes, while it was negative for the other genotypes. Rad_700_1000 did not improve the accuracy in predicting C18:3 compared to modnull; it increased the RMSEP from 0.85 to 1.11 (Table 4). For the 700–100 DDAF period, Rad and Tmin were positively correlated (r = 0.50 and p < 0.001). 3.2.3. A C18:3 model using the minimum temperature and solar radiation We investigated whether the prediction accuracy could be improved using the minimum temperature and solar radiation. Two strategies were tested: We looked for the best combination of time windows for Tmin and Rad, and we combined Rad and Tmin in the same time window through the photothermal quotient (PTQ). A model that used Tmin and Rad slightly improved the fit compared to mod0 that used Tmin only. The best model that used Tmin and Rad (mod4) had an RMSE of 0.4376, while the respective value for mod0 was 0.4721 (Table 4). Fig. 4 shows the effect on the models’ fits of combining Tmin and Rad at different periods. The fitting capacities are highly dependent on the Tmin period while the marginal gain due to the addition of Rad to the model is weak, independent of the period. The model with the best fitting capacity (Fig. 5) used Tmin_680_930 and Rad_100_300 (mod4 in Table 4). The effects of Tmin_680_930 and Rad_100_300 on C18:3 were statistically significant (p < 0.001 and p < 0.01, respectively). The effect of Tmin_680_930 on C18:3 depended on the genotype (p < 0.01), unlike the effect of Rad_100_300. The fixed effects of mod4 were used to predict C18:3 using the Eq. (3) of the model mod4 with the random effect set to zero:

3.2.2. Relation between C18:3 and solar radiation For each tested time window (Table 3), the R2-like index of the associated model is plotted in Fig. 2. It reveals two main periods of strong relations between the amount of solar radiation and C18:3: from 100 to 300 DDAF (Rad_100_300) and from 700 to 1000 DDAF (Rad_700_1000). The first period (Rad_100_300) explained slightly better the variability of C18:3 (R2-like index = 0.934) than the second (Rad_700_1000; R2-like index = 0.926). Nevertheless, considering only one environmental variable, Tmin_680_930 had better fit and predictive capabilities (R2-like index = 0.967) than solar radiation during any of the investigated periods (R2-like index = 0.948 and 0.933 for Rad_100 300 and Rad_700_1000, respectively, Table 4). 3.2.2.1. Solar radiation from 100 to 300 DDAF (Rad_100_300). The likelihood ratio test between a model without Rad (modnull in Table 4) and mod1 indicated that Rad_100_300 had a statistically significant effect on C18:3 (p < 0.001). Moreover, the effect of Rad_100_300 on C18:3 depended on the genotype (p < 0.01). An increase in solar radiation between 100 and 300 DDAF was linked to an increase in C18:3 (Fig. 3). The slopes varied from 0.01373 ( ± 0.003) for Cormoran to 0.00092 ( ± 0.003) for MDS10. In general, the impact of Rad_100_300 on C18:3 was weaker for the HOLL genotypes than for the conventional genotypes (Table 5). Rad_100_300 improved the accuracy of the C18:3 predictions compared to modnull; it decreased

C18:3i = Gi + αGi × Tmin_680_930i + 4.008∙10−4 × Rad_100_300i (3) where C18:3i is the linolenic acid concentration i associated with one triplet (Site × Year × Genotype); Gi is the constant associated with the genotype of observation i in the range 3.23, 12.41; αGi is the coefficient associated with the genotype in the range −0.322 and −0.063; Tmin_680_930i is the daily minimum temperature (°C) associated with the couple (Site × Year)i averaged over the period from 680 to 930°Fig. 2. Goodness of fit index (R2-like index) of the models that explain the linolenic acid concentration (C18:3 in%) as a function of the mean solar radiation during a period defined in degree-days after the onset of flowering (DDAF). The 251 DD period (◆) is the period from 680 to 930°-days after the onset of flowering identified as critical for the impact of the minimum temperatures on C18:3 (Baux et al., 2013). Rad_100_300 is the daily solar radiation (W m−2) averaged over the period from 100 to 300 DDAF; Rad_700_1000 is the daily solar radiation (W m−2) averaged over the period from 700 to 1000 DDAF.

170

Field Crops Research 212 (2017) 165–174

M. Gauthier et al.

Fig. 3. Relation between linolenic acid concentration (C18:3 in%) measured in the field and the daily solar radiation (W m−2) averaged over the period from 100 to 300°-days after flowering (Rad_100_300). The related model is C18:3i = μ + Rad_100_300i + Gi + Rad_100_300i × Gi + (Site × Year)i + ϵ i (R2-like index = 0.95, mod1 in Table 4). C18:3i is the linolenic acid concentration i associated with one triplet (Site × Year × Genotype), μ is the mean population of C18:3; Rad_100_300i is the Rad_100_300 associated with the couple (Site × Year)I, Gi is the genotype of the observation I, (Site × Year)i is considered random, and all the other effects are fixed; ϵ i is the residual error. The lines are drawn with the model coefficients. For each genotype, number of observations are brackets.

days after the onset of flowering; and Rad_100_300i is the daily solar radiation (W m−2) associated with the couple (Site × Year)i averaged over the period from 100 to 300°-days after the onset of flowering; Overall, mod4 improved the prediction of C18:3 in comparison with mod0, and decreased the RMSEP from 0.635 to 0.586. However, this gain occurred only for conventional genotypes (Table 4). C18:3 decrease with Tmin_680_930 was associated with C18:1 increase (p < 0.0001) as well as C18:2 decrease (p < 0.0001). PTQ was also tested to explain the variability in C18:3. The highest R2-like index was found for the period from 400 to 600 DDAF (0.932). The likelihood ratio tests indicated that PTQ_400_600 had a statistically significant effect on C18:3 (p < 0.001) without significant interaction with the genotype (p > 0.05). Higher PTQ_400_600 was associated with lower C18:3 (an increase of 1 W m−2 °C−1 in PTQ decreased C18:3 by 0.0969 percentage points). However, PTQ_400_600 explained less variability than Rad_100_300 and Tmin_680_930 when the R2-like index of the associated models were compared (mod3, mod0, and mod1 in Table 4, respectively). Moreover, PTQ_400_600 (mod3) had less

predictive accuracy than mod4 that considers the minimum temperature and solar radiation additively, at different time windows. 3.2.4. Relations of solar radiation from 100 to 300 DDAF (Rad_100_300) with other quality traits and production traits The effect of Rad_100_300 was further tested on other quality traits (i.e., oil concentration, C18:1 and C18:2), yield and yield components, in order to identify a possible indirect effect of radiation registered during this period, on C18.3 concentration. Significant effects of Rad_100_300 on oil concentration and fatty acids composition were found, as well as significant interactions between these factors and the genotype. Higher Rad_100_300 was associated to higher or lower oil concentration, depending on the genotype. Opposite effects on C18:1 and C18:2 concentration were observed in some genotypes. A negative effect on thousand kernel weight and a positive effect on seed number per square meter was only observed while selecting a subset of data (n = 128) with only the genotypes tested in environments with a large range of solar radiation during the 100–300 DDAF period (Table 6). Fig. 4. R2-like index of the models explaining the linolenic acid concentration (C18:3 in%) as a function of solar radiation (Rad) and the minimum temperature (Tmin) where each variable was averaged during a period defined in degree-days after the onset of flowering (DDAF). Each point is a model with one combination of periods for Tmin and Rad.

171

Field Crops Research 212 (2017) 165–174

M. Gauthier et al.

negative impact of light intensity on C18:2 and C18:3 concentrations and positive on C18:1, consistent with the findings of Izquierdo et al. (2009) with an in-field shading treatment. However in our experiment, the decrease in C18:3 was statistically significant for only one genotype (MSP21, a very low linolenic genotype) and independent of the temperature. The lower radiation level that was achieved in the chambers could have lead all plants to a deficit of radiation and therefore a less clear effect on oil composition. A negative impact of solar radiation on C18:3 was also found in the field, specifically during the 700–1000 DDAF period, with a higher sensitivity for the conventional genotypes than for the HOLL genotypes. However, the lack of predictive capacity of Rad_700_1000 may suggest that the relation with solar radiation was, at least partially, due to the correlation between solar radiation and temperature in natural conditions. In contrast, Wang et al. (2016) observed that C18:1 and C18:3 decreased after the pods were shaded during the seed growth period. These results should be carefully considered as local variations in the irradiance might lead to changes in the microclimate, and especially temperature, known to affect oil composition. Moreover, the genotype studied by these last authors showed increases in erucic acid that suggest a different enzymatic pool compared with other studies. The results from the field data allowed us to identify an earlier sensitive period, from 100 to 300 DDAF, for which an increase in solar radiation was linked to higher C18:3, with a stronger effect for the conventional genotypes than for the HOLL genotypes, and variations in C18:1 and C18:2 highly dependent on the genotypes. For this period, the models also suggested an independent effect of solar radiation and temperature on C18:3, as it was pointed out for C18:1 in sunflower (Echarte et al., 2010). The positive impact of solar radiation on C18:3 during an early period was also suggested by Wang et al. (2016), although they applied the shading treatment at the end of the flowering period while the present study results showed the influence of radiation at an earlier stage. As it was demonstrated for the temperature, the variation of fatty acids composition in response to solar radiation depended on the genotype, with a tendency to be less intense for the HOLL genotypes, but without a clear distinction from the conventional genotypes. The lack of cut-off effects between the two types of genotype could be explained by the large number of genes involved in the control of fatty acid production (Ohlrogge and Jaworski, 1997), most of them shared by both types of genotypes. Moreover, since our study considers incident solar radiation, and not intercepted radiation per plant, more variability could emanate from the variations of canopy architecture with genotypes and environment, leading to a less precise relation between radiation and photosynthetic activity than in previous works (Izquierdo et al., 2009).

Fig. 5. The estimated linolenic acid concentration (C18:3 in%) using the model with the minimum temperature and solar radiation with the random effect set to zero vs. observed index = 0.97). The model equation is C18:3 (R2-like C18:3i = Gi + αGi × Tmin_680_930i + 4.008∙10−4 × Rad_100_300i with C18:3i: C18:3 is associated with one triplet (Site × Year × Genotype), Gi is the constant associated with the genotype of observation I, αGi is the coefficient associated with the genotype; Tmin_680_930i is the daily minimum temperature (°C) associated with the couple (Site × Year)i averaged over the period from 680 to 930°-days after the onset of flowering, and Rad_100_300i is the daily solar radiation (W m−2) associated with the couple (Site × Year)i averaged over the period from 100 to 300°-days after the onset of flowering.

4. Discussion 4.1. Effects of solar radiation and temperature on fatty acid composition A negative impact of temperature on C18:3, associated with an accumulation of C18:1, was confirmed in both field and semi-controlled conditions, C18:3 being the most sensitive to the minimum temperature during the period from 680 to 930 DDAF. These results allowed us to validate Baux et al.’s (2013) findings with a larger dataset including more genotypes and years. The HOLL genotypes were reported to be less sensitive to temperature than the conventional genotypes (Baux et al., 2013). Using more genotypes and environments, our results confirmed a genotypic variability in sensitivity to temperature, as shown in the sunflower by Izquierdo and Aguirrezábal (2008). However, a larger range of sensitivities among the conventional genotypes led to a less clear difference between the conventional and HOLL varieties than that found by Baux et al. (2013). Within the HOLL genotypes, we did not find statistically significantly different sensitivity to temperature. This could be explained by the genotypic similarities of the tested varieties, but also by the fact that they are all characterized by low C18:3 concentration with little variability. The semi-controlled conditions experiment revealed an overall

4.2. Hypotheses regarding the effects of solar radiation on oil composition The positive effect of incident solar radiation from 100 to 300 DDAF

Table 6 Quality and production traits explained by solar radiation averaged over the period from 100 to 300°-days after the onset of flowering (Rad_100_300). Coefficient’s Sign

Oil concentration (%)

C18:1 (%)

C18:2 (%)

C18:3 (%)

Seed yield (t/ha)*

Positive

Hybrirock, Bonanza

MDS10

V316OL, V280OL, Expert

All conventional genotypes and most of the HOLL genotypes Magnitude is genotype-dependent.

NS

Negative

MDS10

V316OL, Expert, Standing, Aviso

MDS10

NS

Thousand-kernel weight (g) *

Number of seeds per m2 * X

X

The coefficient’s sign is the Rad_100_300 coefficient significantly different from zero. The names of the genotypes indicate that the effect of Rad_100_300 is statistically significant (positive or negative impact depending on the row) for these genotypes. Underlining refers to the HOLL genotypes. NS: no statistically significant effect at p < 0.05. X: overall a statistically significant effect (p < 0.05) of Rad_100_300 on the variable without significant interaction with the genotype. *: The relation was shown in a smaller dataset.

172

Field Crops Research 212 (2017) 165–174

M. Gauthier et al.

prediction accuracy of C18:3 before physiological maturity using both minimum temperature and solar radiation. The precision of the prediction was slightly improved by adding the average solar radiation from 100 to 300 DDAF to the model. The results also indicate that combining Rad and Tmin through the PTQ did not improve the fitting or the prediction accuracy of C18:3. The comparison of the different models revealed that solar radiation added only a little information to the model compared to minimum temperature. However, for the prediction of C18:3 concentration at harvest, solar radiation from 100 to 300 DDAF gives earlier insights than the Tmin_680_930 index by about 30 days in Swiss conditions, which could be a valuable information for oil seed crushing industry. A potential application of this model that explains the variations in C18:3 as a function of the genotype, minimum temperature, and solar radiation could provide an ad hoc analysis of the production systems.

on C18:3 concentration occurred during a period when the pods are elongating and the seeds are just initiating. During this early period, the photo-assimilates required for the pod and seed development are supplied by the leaves and are allocated primarily to pod development (Jullien et al., 2011). Leterme (1985) showed also that it is a key period, during which the number of seeds per pod is set to the maximum, if assimilates are available quickly, and in large enough quantity. The present study confirmed the positive relation between solar radiation and the number of seeds per square meter during the 100–300 DDAF phase. Moreover, the thousand-kernel weight linearly decreased with the solar radiation levels suggesting that the increase in assimilate availability during this early period modified the source-sink balance in favor of the sinks. Grain filling occurs during the autotrophic phase (Leterme, 1985). During this phase, the solar radiation levels (Willms et al., 1999) and temperature (Canvin, 1965; Willms et al., 1999) are associated with an increase in the photosynthetic activity and thus increase the source-sink ratio. As described by Echarte et al. (2012) for the sunflower, an increase in assimilates availability during oil synthesis could saturate the desaturase enzymes, as well as the plastid protein membrane transporter, and result in a lower percentage of poly-unsaturated fatty acids. Therefore, higher concentration of C18:3, could be the result of either low radiation during seed filling or high radiation at an earlier stage, which is consistent with the observation of 2 sensitive phases, 100–300 DDAF and 700–1000 DDAF with opposite effects of incident radiation on C18:3. However, due to the autotrophic phase during seed filling (Leterme, 1985), increasing the seed-number (sink) through higher light interception in an early phase of seed development also results in a source increase through the pod length, responsible for most of the assimilation during oil synthesis. With the impact of the solar radiation depending on the period, it is not surprising to find a lesser effect in OSR compared to the sunflower (Echarte et al., 2012), or even contradictory results (Wang et al., 2016). Moreover, these biosynthesis steps have been affected through the breeding of HOLL cultivars, which could explain the lesser extend of C18:3 increase when the sinks (i.e. the radiation at an early stage of seed development) increase, but also the large variability of impacts on the balance between C18:1 and C18:2. Another hypothesis regarding the impact of solar radiation would be a direct effect of radiation on the enzymatic chain, as it is observed for temperature, (Wilmer et al., 1998; Garcia-Diaz et al., 2002). The transcription of FAD3 and FAD8 genes has been reported to decrease in darkness which preceded C18:3 content reduction in soybean (Collados et al., 2006). Similarly, FAD2 was regulated at transcriptional level by light in cotton leaves, which was associated with C18:2 increase (Kargiotidou et al., 2008). This hypothesis supports changes in the fatty acid ratios without affecting the grain yield components, and the differences observed among genotypes, and especially between conventional and HOLL OSR. In summary, it is highly probable that solar radiation impacts the fatty acid composition by mediating the source-sink ratio differently at different growth stages. Due to the succession of heterotrophic and autotrophic phases in OSR yield construction, the impact of solar radiation on the oil composition is variable in OSR with a difference in sensitivity among the genotypes. Additionally, a regulation of the desaturases enzymes by light would bring even more variability among genotypes.

5. Conclusion Temperature and radiation affect the fatty acid composition of OSR, through various mechanisms. The direct regulation of the desaturase enzymes by temperature is well-known, although with large differences among genotypes. Different results concerning the effects of light can be explained by an indirect impact through the accumulation of photoassimilates and the modification of the source-sink ratio depending on the radiative exposure period. Prediction of C18:3 concentration with good accuracy through the temperature, solar radiation, and genotype is possible. Acknowledgements This work was co-funded by the Swiss Confederation’s Innovation Promotion Agency and industrial partners as part of the project Innovative Production Technology for High Oleic Oilseed Rape (Brassica napus) with Very Low Alpha-Linolenic Acid Content (CTI project 12950.1;10 PFLS-LS). References Ascherio, A., Willett, W.C., 1997. Health effects of trans fatty acids. Am. J. Clin. Nutr. 66, 1006–1010. Barta, D., Tibbitts, T., Bula, R., Morrow, R., 1992. Evaluation of light emitting diode characteristics for a space-based plant irradiation source. Adv. Space Res. 12, 141–149. Bates, D., Maechler, M., Bolker, B., Walker, S., 2015. Fitting linear mixed-Effects models using lme4. J. Stat. Software 67, 1–48. Baux, A., Hebeisen, T., Pellet, D., 2008. Effects of minimal temperatures on low-linolenic rapeseed oil fatty-acid composition. Eur. J. Agron. 29, 102–107. Baux, A., Colbach, N., Pellet, D., 2011. Crop management for optimal low-linolenic rapeseed oil production—field experiments and modelling. Eur. J. Agron. 35, 144–153. Baux, A., Colbach, N., Allirand, J.M., Jullien, A., Ney, B., Pellet, D., 2013. Insights into temperature effects on the fatty acid composition of oilseed rape varieties. Eur. J. Agron. 49, 12–19. Bristow, K.L., Campbell, G.S., 1984. On the relationship between incoming solar radiation and daily maximum and minimum temperature. Agric. Forest Meteorol. 31, 159–166. Byrnes, J., 2008. Re: Coefficient of Determination (R^2) when Using Lme. http://thread. gmane.org/gmane.comp.lang.r.lme4.devel/684. Canvin, D.T., 1965. The effect of temperature on the oil content and fatty acid composition of the oils from several oil seed crops. Can. J. Botany 43, 63–69. Carré, P., Evrard, J., Judde, A., Labalette, F., Mazette, S., 2007. Technological performances of low linolenic/high oleic rapeseed oils for food and non-food application. The 12th International Rapeseed Congress 152–159. Champolivier, L., Merrien, A., 1996. Effects of water stress applied at different growth stages to Brassica napus L. var oleifera on yield, yield components and seed quality. Eur. J. Agron. 5, 153–160. Collados, R., Andreu, V., Picorel, R., Alfonso, M., 2006. A light-sensitive mechanism differently regulates transcription and transcript stability of omega3 fatty-acid desaturases (FAD3, FAD7 and FAD8) in soybean photosynthetic cell suspensions. FEBS Lett. 580, 4934–4940. Deng, X., Scarth, R., 1998. Temperature effects on fatty acid composition during development of low-linolenic oilseed rape (Brassica napus L.). J. Am. Oil Chem. Soc. 75, 759–766. Diepenbrock, W., 2000. Yield analysis of winter oilseed rape (Brassica napus L.): A review. Field Crops Res. 67, 35–49. Durruty, I., Aguirrezabal, L.A.N., Echarte, M.M., 2016. Kinetic modeling of sunflower

4.3. A linolenic acid concentration model Baux et al. (2013) proposed an empirical model to predict C18:3 in winter OSR using the average minimum temperature from 680 to 930 DDAF. This model was validated with extensive field data. It improved mainly the prediction capacity for conventional genotypes as the impact of temperature is higher for these genotypes than for HOLL genotypes. We then investigated whether it was possible to improve the 173

Field Crops Research 212 (2017) 165–174

M. Gauthier et al. grain filling and fatty acid biosynthesis. Front. Plant Sci. 7. Echarte, M.M., Angeloni, P., Jaimes, F., Tognetti, J., Izquierdo, N.G., Valentinuz, O., Aguirrezábal, L.A.N., 2010. Night temperature and intercepted solar radiation additively contribute to oleic acid percentage in sunflower oil. Field Crops Res. 119, 27–35. Echarte, M.M., Alberdi, I., Aguirrezábal, L.A.N., 2012. Post-flowering assimilate availability regulates oil fatty acid composition in sunflower grains. Crop Sci. 52, 818–829. Echarte, M.M., Puntel, L.A., Aguirrezabal, L.A.N., 2013. Assessment of the critical period for the effect of intercepted solar radiation on sunflower oil fatty acid composition. Field Crops Res. 149, 213–222. Département fédéral de l’intérieur suisse, 2008. Ordonnance du DFI sur les huiles et graisses comestibles et leurs dérivés, Modification du 7 mars 2008. RS 817.022.105. Garcia-Diaz, M.T., Martinez-Rivas, J.M., Mancha, M., 2002. Temperature and oxygen regulation of oleate desaturation in developing sunflower (Helianthus annuus) seeds. Physiol. Plant. 114, 13–20. Habekotté, B., 1997. A model of the phenological development of winter oilseed rape (Brassica napus L.). Field Crops Res. 54, 127–136. Izquierdo, N.G., Aguirrezábal, L.A.N., 2008. Genetic variability in the response of fatty acid composition to minimum night temperature during grain filling in sunflower. Field Crops Res. 106, 116–125. Izquierdo, N., Aguirrezábal, L., Andrade, F., Pereyra, V., 2002. Night temperature affects fatty acid composition in sunflower oil depending on the hybrid and the phenological stage. Field Crops Res. 77, 115–126. Izquierdo, N.G., Aguirrezábal, L.A.N., Andrade, F.H., Geroudet, C., Valentinuz, O., Pereyra Iraola, M., 2009. Intercepted solar radiation affects oil fatty acid composition in crop species. Field Crops Res. 114, 66–74. Jullien, A., Mathieu, A., Allirand, J.M., Pinet, A., de Reffye, P., Cournede, P.H., Ney, B., 2011. Characterization of the interactions between architecture and source-sink relationships in winter oilseed rape (Brassica napus) using the GreenLab model. Ann. Bot. 107, 765–779. Kargiotidou, A., Deli, D., Galanopoulou, D., Tsaftaris, A., Farmaki, T., 2008. Low temperature and light regulate delta 12 fatty acid desaturases (FAD2) at transcriptional level in cotton (Gossypium hirsutum). J. Exp. Bot. 59, 2043–2056. Koletzko, B., Decsi, T., 1997. Metabolic aspects of trans fatty acids. Clin. Nutr. 16, 229–237. Lancashire, P.D., Bleiholder, H., Boom, T.V.D., LangelÜDdeke, P., Stauss, R., Weber, E., Witzenberger, A., 1991. A uniform decimal code for growth stages of crops and weeds. Ann. Appl. Biol. 119, 561–601. Landschoot, S., Waegeman, W., Audenaert, K., Vandepitte, J., Haesaert, G., De Baets, B.,

2012. Toward a reliable evaluation of forecasting systems for plant diseases: a case study using fusarium head blight of wheat. Plant Disease 96, 889–896. Lenth, R.V., 2016. Least-Squares Means: The R Package Lsmeans. pp. 33 69. Leterme, P., 1985. Modélisation de la croissance et de la production des siliques chez le Colza d'hiver (. Doctoral dissertation. Matthäus, B., 2007. High oleic low linolenic rapeseed oil as alternative to common used frying oils. In: the 12th International Rapeseed Congress on Qulity, Nutrition and Processing. Wuhan, China. pp. 165–168. Mollers, C., 2002. Development of high oleic acid oilseed rape. Presentation at the 8th International Conference for Renewable Resources and Plant Biotechnology. Narossa R 10–11. Ohlrogge, J.B., Browse, J., 1995. Lipid biosynthesis. Plant Cell 7, 957–970. Ohlrogge, J.B., Jaworski, J.G., 1997. Regulation of fatty acid synthesis. Annu. Rev. Plant Biol. 109–136. Prieto, J.I., Martínez-García, J.C., García, D., 2009. Correlation between global solar irradiation and air temperature in Asturias, Spain. Sol. Energy 83, 1076–1085. Pritchard, F.M., Eagles, H.A., Norton, R.M., Salisbury, P.A., Nicolas, M., 2000. Environmental effects on seed composition of Victorian canola. Aust. J. Exp. Agric. 40, 679–685. R Core Team, 2015. R: A Language and Environment for Statistical Computing. R Foundation for Statistical Computing, Vienna, Austria. Singer, S.D., Zou, J., Weselake, R.J., 2016. Abiotic factors influence plant storage lipid accumulation and composition. Plant Sci. 243, 1–9. Trémolières, A., Dubacq, J.P., Drapier, D., 1982. Unsaturated fatty acids in maturing seeds of sunflower and rape: regulation by temperature and light intensity. Phytochemistry 21, 41–45. Wang, C., Hai, J., Yang, J., Tian, J., Chen, W., Chen, T., Luo, H., Wang, H., 2016. Influence of leaf and silique photosynthesis on seeds yield and seeds oil quality of oilseed rape (Brassica napus L.). Eur. J. Agron. 74, 112–118. Weymann, W., Böttcher, U., Sieling, K., Kage, H., 2015. Effects of weather conditions during different growth phases on yield formation of winter oilseed rape. Field Crops Res. 173, 41–48. Willms, J.R., Salon, C., Layzell, D.B., 1999. Evidence for light-Stimulated fatty acid synthesis in soybean fruit. Plant Physiol. 120, 1117–1128. Wilmer, J.A., Lessire, R., Helsper, J.P.F.G., Van Der Plas, L.H.W., 1998. Regulation of elongase activity by abscisic acid and temperature in microspore-derived embryos of oilseed rape (Brassica napus). Physiol. Plant. 102, 185–191. Zuil, S.G., Izquierdo, N.G., Luján, J., Cantarero, M., Aguirrezábal, L.A.N., 2012. Oil quality of maize and soybean genotypes with increased oleic acid percentage as affected by intercepted solar radiation and temperature. Field Crops Res. 127, 203–214.

174