The E ﬀ ect of Climate Change on Linolenic Fatty Acid in Oilseed Rape

: Linolenic acid content plays a signiﬁcant role in the quality of oilseed rape oil. With the use of a model developed to calculate linolenic acid content depending on temperature, we performed simulations focused on the predicted changes of linolenic acid content triggered by expected climate change. The aim of the study was realized with the use of meteorological data recorded at 16 locations, representing 16 provinces in Poland, in the period of 1986–2005, and values obtained after the transformation of the recorded data to reﬂect temperature changes in four di ﬀ erent representative concentration pathway scenarios (RCP2.6, RCP4.5, RCP6.0, RCP8.5) according to the gfdl_esm2m climate model. The results of the simulations performed under all tested scenarios (RCP2.6, RCP4.5, RCP6.0, RCP8.5) for all analyzed periods (2020–2039, 2040–2059, 2060–2079, 2080–2099) indicate a reduction of linolenic acid content as an e ﬀ ect of climate change. The comparison of simulation results obtained with and without allowing for the inﬂuence of temperature on oilseed rape development revealed the dominant role of direct inﬂuence in modifying the linolenic acid content. An analysis of the results also indicated how indirect inﬂuence of climate change, as expressed in changes of linolenic acid content generated by the acceleration of oilseed rape maturation, reduces the direct inﬂuence of climate change, expressed in changes of linolenic acid content caused by temperature. Additionally, simulation results point to opportunities, triggered by climate change, for a diversity of regions in terms of linolenic acid content in comparison to the years of 1986–2005.


Introduction
Oilseed rape is one of the most important oilseed crops in the world [1][2][3]. The major producers of this crop are China, India, Canada, and the European Union [4]. World production of oilseed rape is increasing very rapidly [5][6][7][8]. In Poland, due to increased demand for raw materials for biofuels, production of oilseed rape cultivation has increased by almost double over the last two decades, from 466 ha in 1998 to 823 thousand ha in 2017 [9]. Oilseed rape oil is used for various purposes, mainly human nutrition and as a raw material in the chemical industry. The residues from oil production are used as valuable animal forage [10]. The oil content of oilseed rape varies from 45% to 50% [11] depending on fertilization, variety, and meteorological conditions, and temperature plays a leading role in the modification of oilseed rape quantity and the balance between saturated, mono-unsaturated, and polyunsaturated fatty acids [12]. The effects of temperature on fatty acid content have been studied previously. Many reports indicate that cold and heat can affect fatty acid composition in oilseed rape and especially linolenic acid content [13][14][15][16][17][18]. Hence, it can be assumed that there is a high probability that climate change will also alter the content of fatty acids. Unfortunately, Lisek Bazyl Table 2. Models used for predicting oilseed rape maturation.

Selection of A Model for Simulating Oilseed Rape Maturation
The selection of the model for simulating oilseed rape maturation was made using a Taylor diagram [29]. This diagram provides a visual framework to compare the results of the models to the observations. The simulated and observed dates of the beginning of maturation (BBCH 81) of oilseed rape cultivars listed in Table 3 were compared. The simulations were performed using meteorological data registered from March to July at Winna Góra from 2012-2016. Observation results came from field experiments conducted at the same location and time with the use of cultivars listed in the Table 3.

Simulation of the Effect of Climate Change on Linolenic Acid Content
The simulations were performed using both real and virtual meteorological data. The real data was from the period from 1986 to 2005 at 16 locations in Poland (Figure 1), while the virtual data was obtained after the transformation of the recorded data to reflect temperature changes under RCP2.6 RCP4.5 RCP6.0, and RCP8.5 scenarios according to the 16 climate model presented on the Climate Change Knowledge Portal created by the World Bank (http://sdwebx.worldbank.org/climateportal/) ( Table 4). Data transformation was based on the delta change approach, which consists of adding the time mean temperature change between the baseline and scenario simulations to each daily value in the observed time series according to the Equation (1):  Table 4). Data transformation was based on the delta change approach, which consists of adding the time mean temperature change between the baseline and scenario simulations to each daily value in the observed time series according to the Equation (1): where: Tdb-debiased temperature; Tobsref-daily temperature observed in the reference period; Tsimproj-monthly temperature simulated for the projected period; Tsimref-monthly temperature simulated for the reference period.   Two kinds of simulations were carried out in the study. The first generated results allowed for both the direct and indirect influence of temperature on linolenic acid content. The second set of results only allowed for the direct influence of temperature. The direct influence is based on the relation between temperature and linolenic acid content, while indirect influence recognizes the influence of temperature on the rate of oilseed rape development.
The difference between the two simulations was the way in which the date of the beginning of oilseed rape maturation was determined.
In the first case, the sub-model simulating oilseed rape development was used both for the period of 1986-2005 as well as for the periods of 2020-2039, 2040-2059, 2060-2079, 2080-2099. In the second simulation, the dates of oilseed rape maturation indicated for the period of 1986-2005 were also used in simulations for the other periods.
For each location included in the study, 340 (20 years × 17 sets of meteorological data) simulations were performed to assess linolenic acid content.
where Lac is linoleic acid content; T 20 is the average temperature ( • C) of the twenty days from the beginning of oilseed maturation; and a, b the equation coefficients (Table 5). Equation (2) was developed on the basis of the data acquired by Wójtowicz [30] in  Table 6. High R 2 values, equal to 0.87, ensure the predictive model fits the data well. The direct influence of temperature on linoleic acid content was calculated using Equation (3).
where ID is the direct impact of climate change on linoleic acid content; Lac-tor i, j is the linoleic acid content obtained from simulations without allowing for the influence of temperature on development of oilseed rape for the i scenario and j period (i: RCP2.6, RCP4.5, RCP6.0, RCP8.5; j: 2020-2039, The indirect impact of climate change on linoleic acid content was calculated using Equation (5).
where II is the indirect impact of climate change on linoleic acid content; I(D + I) is the sum of the direct and indirect impact of climate change on linoleic acid content; and ID is the direct impact of climate change on linoleic acid content. Additionally, with the use of the Euclidean distance between each pair of the linolenic acid content obtained in simulations allowing for both direct and indirect influence of climate change on linolenic acid content, we performed a territorial division for regions that share similar values based on complete (or maximum) clustering methodology. This method computes all pairwise dissimilarities between the elements in cluster 1 and the elements in cluster 2, and considers the largest value (i.e., maximum value) of these dissimilarities as the distance between the two clusters. The complete clustering algorithm gave compact clusters that were then split with the threshold value of Euclidean distance equaling 0.2, as shown in Figure 2.

The Development of Models Predicting the Beginning of Oilseed Rape Maturation
The maturation of oilseed rape was dependent on the year of the study, but not on the cultivar used.
In the experiments conducted in the period of 2007-2011, the earliest symptoms of oilseed rape maturation was noticed in 2008 (9.06), while the latest was in 2010 (25.06) ( Table 7). Six temperature-response models and four combinations of the models composed of two functions (monomolecular and exponential, Gompertz and exponential, logistic + exponential) were used to express the relationship between oilseed rape maturation and accumulated degree days. The values of the coefficients of the models and statistical values for the models used are summarized in Table 8.

Selection of a Model for Simulating Oilseed Rape Maturation
Similar to the experiments aimed at the model development, no differences in the beginning of maturation between cultivars used in the years of 2012-2016 were observed. Again, maturity of oilseed rape was triggered by meteorological conditions and started between the 2nd and 26th of June (Table 9). The differences between the observation and simulation results depending on the year, and the model ranged between 0 and 22 days (Table 10). The results of the m+exp model and the mean values of results of the m+exp and G+exp models, as well as the m+exp and lo+exp models, did not differ with observation by more than 9 days. The comparison of the G+exp model and the mean values of the three models (m+exp and G+exp and lo+exp) with the results of the observations revealed that the differences did not exceeded 10 days, whereas the maximum disagreement between the mean values of the results of the G+exp and lo+exp models and observations was 11 days. A one-day maximum disagreement between the model results and real data appeared as an effect of the use of the lo+exp model. The other models generated results which differed with observations by 18 (m), 20 (G), and 22 (lo) days.
The use of a Taylor diagram revealed that combining the results of the two models (m+exp and lo+exp) appeared on the plot at the smallest distance from the observation point ( Figure 3). Based on the results obtained using the diagram, this combination of models was selected for further analysis.
Agronomy 2020, 10, x FOR PEER REVIEW 10 of 21 The use of a Taylor diagram revealed that combining the results of the two models (m+exp and lo+exp) appeared on the plot at the smallest distance from the observation point ( Figure 3). Based on the results obtained using the diagram, this combination of models was selected for further analysis.

Simulation of the Effect of Climate Change on the Beginning of Oilseed Rape Maturation
A simulation based on data registered in 16 analyzed locations performed with the model obtained by combining the two model results (m + exp and lo + exp) revealed that oilseed rape maturation started between 162 and 179 days after the 1st of January (Table 11). For the period of 2020-2039, depending on the RCP scenario, the simulation showed a reduction of the time needed for the oilseed rape to start maturation by 12 Table 10. Comparison of the observed beginning of oilseed rape maturation and simulated with the use of models listed in Table 2.

Simulation of the Effect of Climate Change on the Beginning of Oilseed Rape Maturation
A simulation based on data registered in 16 analyzed locations performed with the model obtained by combining the two model results (m + exp and lo + exp) revealed that oilseed rape maturation started between 162 and 179 days after the 1st of January (Table 11). For the period of 2020-2039, depending on the RCP scenario, the simulation showed a reduction of the time needed for the oilseed rape to start maturation by 12
The coefficient of variation (CV) of linolenic acid content simulated for the period of 2020-2039 under four RCP scenarios varied between 0.24% and 0.69% (Table 12). For 75% of locations (twelve out of sixteen), the analyzed parameter was smaller than 0.46. The minimum and maximum values of the coefficient of variation for the period of 2040-2059 were 0.40% and 0.92%. The analyzed parameter was smaller than 0.80 for 75% of locations. For the period of 2060-2079, the coefficient of variation (CV) varied between 0.75% and 1.60%. For twelve out of sixteen locations (75%), the analyzed parameter was smaller than 1.42. Simulations for the period of 2080-2099 revealed that the value of the parameter ranged between 1.00% and 2.01%, and for 75% of locations was smaller than 1.58.
Simulations which didn't allow for the influence of climate change on the rate of oilseed rape development performed for the period of 2020-2039 revealed that the minimum and maximum content of the linolenic acid was 9.27 and 9.75 for RCP2.6, 9.29 and 9.76 for RCP4.5, 9.30 and 9.81 for RCP6.0, and 9.18 and 9.60 for RCP8.5 (Table 13) Comparisons of linolenic acid content obtained in simulations performed with and without allowing for the influence of climate change on the rate of oilseed rape development revealed that under the RCP2.6 scenario, the former generated higher changes in the analyzed parameters irrespective of the analyzed period. (Table 14)   Under the scenarios of RCP2.6 for the periods of 2060-2079, the locations are grouped into two clusters (Figure 4). Four clusters were distinguished under the RCO6.0 scenario for the periods of 2060-2079 and 2080-2099, and under RCP8.5 scenario for the period of 2060-2079, while for the other combinations of scenario and period, the locations were grouped into three clusters. Table 14. Changes of linolenic acid content predicted for four scenarios (RCP2.6, RCP4.5, RCP6.0, RCP8.5) and four periods (2020-2039, 2040-2059, 2060-2079, 2080-2099) influenced by climate change.

Discussion
The agreement between the results of observations and the simulations based on mean results of two models (m + exp and lo + exp) indicates that the equation developed for predicting the onset of the beginning of oilseed rape maturity (BBCH 81) can be applied as a sub-model in simulations aimed at prediction of linolenic acid content in oilseed rape seeds under climate change scenarios. The comparison of the clustered results revealed that Gdańsk (16 times) and Szczecin (14 times) appeared most frequently in the cluster with the lowest acid content. Kraków (16 times), Opole (16 times), Rzeszów (16 times), and Warszawa (16 times) were most often classified into the cluster with the highest acid content.

Discussion
The agreement between the results of observations and the simulations based on mean results of two models (m + exp and lo + exp) indicates that the equation developed for predicting the onset of the beginning of oilseed rape maturity (BBCH 81) can be applied as a sub-model in simulations aimed at prediction of linolenic acid content in oilseed rape seeds under climate change scenarios.
We described the relationship between linolenic acid content and temperature using a linear regression equation, which corroborates the approach adopted earlier in [18] to estimate the effect of temperature on fatty acids composition in conventional cultivars of oilseed rape in Switzerland. Discussing the results, the authors of that study stressed that linear regression being characterized only by two parameters (slope and intercept) minimizes errors due to parameter estimation. The equation presented in [18] explains approximately half of the variability in linolenic acid content, which did not differ from findings reported in our study. The results obtained could be improved by including more genotypes, years, and locations. However, this deficiency was compensated for, to some extent, by the great number of plots.
The results of the simulations performed under all tested scenarios (RCP2.6, RCP4.5, RCP6.0, RCP8.5) for all analyzed periods (2020-2039, 2040-2059, 2060-2079, 2080-2099) indicate a reduction of linolenic acid content as an effect of climate change. These findings corroborate the results of Namazkar et al. [32,33], who reported a reduction of linolenic acid content in seeds of oilseed rape grown in future climate scenarios with an elevated temperature. The changes of fatty acids composition are induced by variations in temperature during seed development [15]. Conventional oilseed rape oil contains approximately 6% saturated fatty acids (C16:0 + C18:0), 55-60% oleic acid (C18:1), 20-26% linoleic acid (C18:2), and 8-10% linolenic acid (C18:3). Of the fatty acids of oilseed rape, oleic acid (C18:1) and linolenic acid (C18:3) are most sensitive to temperature changes [17]. Higher temperatures during seed maturation are known to be conductive for the formation of more oleic acid, and lower temperatures for formation of linolenic acid, which is the product of linoleic acid desaturation. The final amount of linolenic acid depends on the desaturation rate of C18:1 to C18:2, and then of C18:2 to C18:3. The negative correlation of oleic acid with linolenic acid presented in [30] corroborates the close relationship between unsaturated fatty acids content and indirectly explains the role of desaturase in shaping the final fatty acids composition.
A similar process is observed during sunflower seeds development [34]. Since sunflower does not contain linolenic acid, low temperatures result in higher linoleic acid content. The increase in linoleic acid at low temperatures may be attributed to an enhanced synthesis or activity of the enzyme oleyl-CoA desaturase, which is responsible for converting oleic acid to linoleic acid [35].
The negative effects of elevated temperature on linolenic acid content in seeds of oilseed rape presented in our study corroborate the findings of Deng and Scarth [15], who analyzed fatty acid composition in three temperature regimes and stated that the highest linolenic acid content occurred under the lowest temperature treatment, whereas the lowest was under the highest temperature treatment. This is also consistent with results obtained by Tremolieres et al. [14], who showed an increase in linolenic acid content in mature seed oil as an effect of exposure of oilseed rape plants to low temperatures (12-17 • C). The effect of temperature on the fatty acid composition was also studied in seeds of soybean [36,37] and flax [38,39]. These studies also showed a decrease of fatty acids desaturation when the crops were grown at higher temperatures. Moreover, in soybean seeds [37], a close relationship between unsaturated fatty acids content was confirmed by the significant negative correlation of oleic acid content with the content of linoleic and linolenic acids.
Interesting results of a study focused on environmental effects on oil quality of oilseed crops are also presented by Roche et al. [40], who stated differentiating effects of climatic conditions not only on fatty acids compositions, but also on phytosterols content in safflower seeds. According to Belo et al. [41], the impact of environment on phytosterols depends on the plant species. Changes of these bioactive compounds' concentration due to variations of temperature were found in soybean and sunflower seeds [42,43], while no temperature effect on phytosterols has been reported in canola seeds [44].
The comparison of simulation results obtained with and without allowing for the influence of temperature on oilseed rape development presented in our study revealed that global warming will increase the significance of the direct influence on the modification of the linolenic acid content. An analysis of the results obtained indicates that the indirect influence of climate change expressed in changes of linolenic acid content generated by the acceleration of oilseed rape maturation reduces the direct influence of climate change expressed as changes of linolenic acid content caused by temperature. The importance of indirect influence in modifying the linolenic acid content was particularly evident in the results of simulations performed under the RCP8.5 scenario. In this case, linolenic acid content obtained in simulations without allowing for the influence of climate change on oilseed rape maturation was reduced from 9.34 for the period of 2020-2039 to 8.03 for 2080-2099; whereas linolenic acid content obtained in simulations performed allowing for the effect of climate change on oilseed rape for the same periods were 9.89 and 9.59, respectively.
The small values of coefficient of variance (CV) obtained in the comparisons of linolenic acid content under the four RCP scenarios for all studied locations proves the high credibility of results obtained in simulations for all the analyzed periods. However, changes in the value of CV over the course of time indicates differences in the credibility of predictions. The smallest value of CV for the period of 2020-2039 (0.38), and the highest for the period of 2080-2099 (1.31), indicate that the forecasts for these periods are the most and the least reliable, respectively; whereas predictions for the periods of 2040-2059 (0.64) and 2060-2079 (1.11) seem to be of medium credibility.
Results of clustering indicate that the predicted climate changes will cause an increase in the cultivation area of oilseed rape characterized by a reduced content of linolenic acid. This is particularly clearly indicated by the increase in the number of locations included in the cluster with a lower linolenic acid content recorded when comparing the periods of 2020-2039 and 2080-2099 under the RCP6.0 and RCP 8.5 scenarios. However, the simulations performed indicated also locations included most often in the same cluster. Regions with favorable conditions for the production of rape with the highest (Gdańsk, Szczecin) and the lowest (Kraków, Opole, Rzeszów, Warsaw) content of linolenic acid were distinguished this way. The former would allow for the harvesting of oilseed rape seeds with acid content most similar to that obtained so far. Meteorological conditions in the latter regions would be conducive to the production of oilseed rape aimed at obtaining oil with greater durability, which is important both as a raw material intended for human consumption and for the production of biofuels. Therefore, the results of the simulations presented indicate that the future climate change will not negatively affect the bio-fuel quality. On the other hand, linolenic acid not being synthesized by humans is preferred in a human diet, since it is assume to reduce risk of cardiovascular diseases. Therefore, decrease of its content in the oil of oilseed rape may degrade the quality of oil for consumption and pose a serious challenge for the oil producers in the future. It must be stressed that temperature is not the only parameter which affects the linolenic acid content. Elevation of CO 2 and O 3 and may also trigger the changes in the oil quality. Moreover, the model used in our study does not include characteristics related to genotypic variability of the future cultivars. However, comparison of results presented in [30] showed that variability of linolenic acid content due to differences in temperature between the years of the experiment was approximately ten times bigger than that caused by differences in genotypes of the cultivars used. This is consistent with findings presented in [18] by Baux et al., who for most oilseed rape cultivars tested in the study also observed a greater effect of climate change than genotypic variability on linolenic acid content. Therefore, according to these authors, the potential of the year and location for changing linolenic acid content in oilseed rape seeds can be estimated using average regression parameters.
However additional work aimed at determining not only the effect of cultivar variability, but also other factors like radiation and drought on fatty acids composition, is necessary.