In the study, commercially important varities such as the Tombul variety, which is also known as ‘Yağlı fındık’ by the local people, Kara and Sivri varieties were analyzed. Specimens were harvested from hazelnut groves in Akköy İkisu, the Batlama valley of the Giresun province (Turkey) in August. The geographical coordinates of the grove where the samples were taken are 40 degrees 51 minutes 38.85 seconds North, 38 degrees 18 minutes 56.66 seconds East. Harvested hazelnuts were sun-dried on the concrete threshing floor, on jute matting in accordance with completely traditional methods. Each of the dried hazelnut varieties was completely filled into 80 kg jute sacks and stored in a warehouse whose temperature and humidity cannot be externally adjusted for approximately one year until the next harvest time. During the storage period, the temperature of the warehouse environment ranged from 7.9 to 24.8 °C and the relative humidity ranged from 47.5 to 79.4%. The sacks were placed on wooden planks in order to prevent contact with the concrete floor. The first specimens were taken at the end of the blending process in September and the others were taken in November, January, March, May and August. Samples were taken from the sacks and placed in plastic bags where they were kept at -20 ºC in a deep freezer until analysis.

The analytical-grade chemical substances used in the study were: hydrochloric acid, methanol, ethyl alcohol, chloroform, calcium chloride, iso-octane, potassium hydroxide, diethyl ether, sodium hydroxide, potassium iodide, acetic acid, sodium thiosulfate and a methyl orange indicator, all provided by Merck (Germany), Sigma (ABD), Carlo Erba (France) firms. 37 fatty acid methyl ester mixtures were supplied by Bellefonte PA (USA).

Hazelnut oil extraction. The extraction of lipids from hazelnut samples was done according to the method used by Evren (2011) (Bligh and Dyer, 1959). After the addition of 50 ml of chloroform: methanol (GC purity) (2:1 v/v) to 10 g of sample, 10 ml of 0.01 M CaCl₂ solution were added dropwise to the resulting mixture. The solution was filtered through ordinary filter paper. The resulting filtrate was placed in teflon tubes and centrifuged at 2000 g for 15 minutes. The bottom phase of the centrifuge tube was transferred to a socket volumetric flask using a pipette and the solvent phase was separated on a rotary evaporator to obtain oil. In order to obtain fatty acid methyl esters (ISO, 1978), 0.5 g of oil was weighed into the flask and 4 ml of iso-octane and 2 ml of methanolic KOH solutions were added and then shaken for 30 seconds. The bottom phase of the centrifuge tube was transferred to a socket volumetric flask using a pipette and the solvent phase was separated off on a rotary evaporator so the oil was obtained.

Fatty acid composition. To obtain the fatty acid methyl esters (ISO, 1978), 0.5 g of oil was weighed into an erlenmayer flask and mixed with 4 ml of iso-octane and 2 ml of a methanolic KOH solution, and shaken for 30 seconds. Then, the erlenmayer was closed and left in the dark for 6 minutes. At the end of the period, 2 drops of 1% methyl orange indicator were added and the solution was titrated with 1M HCl until a pink color appeared. After the contents rested for 15 minutes, the colorless upper layer was transfered into glass vials and analyzed in GC. The composition of fatty acids was determined using Shimadzu brand (Model GC-2010, Japan) gas chromatography with a flame ionization detector (FID) and TR-CN100 column (60 m x 0.25 mm I.D., 0.20 μm) (Teknokroma, Spain). The injector and detector temperatures were set at 250 °C. The amount of sample injected was 1.0 μl and helium was used as carrier gas at a pressure of 200 kPa. Injection was performed at a ratio of 1:100. The column temperature was held at 90 °C for 7 minutes, then increased to 240 °C at 5 °C/min. Finally, it was held at 240 °C for 15 minutes. Fatty acids were identified by comparison with the time of arrival of the FAME mixture (Supelco 37 Component FAME Mixture, Cat. No. 18919-1AMP, Bellefonte PA, USA) consisting of 37 standard components (Saricaoglu and Turhan, 2013).

Free fatty acids. Two handfuls of hazelnut sample were put in a cloth bag and hazelnut oil was obtained with the help of a cold press. This oil was placed in a 10 ml centrifuge tube and the sediments in the oil were allowed to settle. About 5 g of fat were transferred to a 250 ml Erlenmeyer flask with a known tare weight. 50 ml of a mixture of ether:ethanol (1/1, v/v) was added and shaken, 2-3 drops of 1% phenolphthalein indicator were dropped and the mixture was titrated with 0.1 N NaOH solution until a light pink color was obtained. The amount of free fatty acid in oleic acid was calculated by the following formula: % Free Fatty Acid = [(SxFx0.0282)/M] x 100, S: Amount of NaOH spent in the titration, F: Factor of 0.1 N NaOH solution, M: Amount of oil used (AOCS, 1990).

Peroxide value. Two handfuls of hazelnut sample were put in a cloth bag and the hazelnut oil was obtained with the help of cold press. About 5-5.09 g oil were weighed into a 250 ml erlenmeyer which was tared and wrapped with aluminum foil. 30 ml of an acetic acid:chloroform (3:2, v/v) mixture were added and mixed. 0.5 ml of a saturated potassium iodide solution was added. It was shaken for a minute and left in the dark. At the end of the shaking period, 30 ml of distilled water and 0.5 ml of 1% starch solution were added. The solution was titrated with 0.01 N sodium thiosulfate solution until a clear cream color was formed and the peroxide number was calculated according to the following formula: Peroxide Number (meq O₂/kg fat) = [(SxNxF)/M] x1000, S: the amount of sodium thiosulfate spent in the titration, F: the factor of the solution, N: the normality of the sodium thiosulfate solution, M: the amount of fat (g) (AOAC, 1990).

Iodine value and O/L. Oxidative stability was calculated with a specific formula given as iodine value (palmitoleic acid * 1.901) + (oleic acid * 0.899) + (linoleic acid * 1.814) + (linolenic acid * 2.737) and expressed as oleic acid/linoleic acid (Belvisio et al., 2017).

For the analysis of data, one-way ANOVA for time and variety, two-way ANOVA to determine time and variety interaction, and the Tukey HSD test to determine different averages were used. The analytical results were evaluated statistically using the SPSS 21 statistical software (SPSS Inc., Chicago, USA).

The changes in fatty acid composition of hazelnuts during storage are given in Table 1. In all hazelnut varieties, the dominant saturated fatty acid was palmitic acid followed by stearic acid. In all varieties, the highest unsaturated fatty acid was identified as oleic acid, followed by linoleic acid. Similar findings were reported in the studies of Oliveira et al., (2008), Alasalvar et al., (2010), Göncüoğlu Taş and Gökmen (2015), Kalkisim et al., (2016).

At the beginning of storage, the highest amount of palmitic acid was found in Sivri (5.60%), the lowest in Tombul (4.63%) and the difference between varieties was found to be significant (p ˂ 0.001). The amount of palmitic acid, which was initially 4.63%, increased to 4.71 (p ˂ 0.01) at the end of storage, but there was no change in other varieties. It was determined that the amount of palmitic acid decreased from 5.74% to 5.35% (Ghirardello et al., 2013) in hazelnuts stored uncontrolled for 12 months but there was no change in another study carried out for 12 months (Koyuncu, 2004). No changes were detected during the storage of hazelnuts packed in a vacuum and stored for one year (Koyuncu et al., 2005). Stearic acid was found to be the 4th most abundant fatty acid in all varieties, while the highest value was determined in Tombul as 2.55% (p ˂ 0.001). Stearic acid increased to 2.71% in Tombul during storage and decreased to 2.11% from 2.27% in Sivri and unchanged in Kara (2.14%). In one study, it was determined that the content of stearic acid in hazelnuts increased from 1.85% to 3.12% as a result of storage for 12 months (Ghirardello et al., 2013) but it did not change in another study (Koyuncu, 2004). Another study carried out on the varieties of Tombul, Palaz and Kalın Kara found no change (Koyuncu et al., 2005). At the beginning and also at the end of storage the highest amount of oleic acid was found in Sivri (84.32%) and the lowest amount was found in Kara (81.06%) (p ˂ 0.001) and the oleic acid amounts of all varieties increased during the storage period (p ˂ 0.001). The increase is thought to be proportional due to the reduction in linoleic acid.

The highest amount of linoleic acid was detected in Kara while the lowest amount was determined in Sivri (p ˂ 0.001). At the end of the storage period, the amount of linoleic acid decreased from 11.46% to 11.00% in Kara, from 8.67% to 8.02% in Tombul and from 7.28% to 6.68% in Sivri. Linoleic acid was also found to decrease in some studies where storage was carried out for 12 months (Koyuncu, 2004; Koyuncu et al., 2005). This decrease is thought to be caused by peroxidation of linoleic acid (Ghirardello et al., 2013).

Saturated fatty acid (SFA) quantities did not change in all varieties at the end of storage (p ˃ 0.05), but the amounts of mono unsaturated fatty acid (MUFA) increased in all varieties, while poly unsaturated fatty acid (PUFA) decreased (Table 2). This is thought to be due to a decrease in the amount of linoleic acid and an increase in the amount of oleic acid at the end of storage. The highest amount of SFA and MUFA among the varieties was found in Sivri (7.98% -84.54% respectively) and the highest amount of PUFA was found in Kara. In one study, the amounts of the SFA, MUFA and PUFA in hazelnuts were found to be 9.75%, 83.70% and 6.53%, respectively (Belviso et al., 2017). According to Koyuncu (2004), in hazelnuts stored for 12 months, the SFA increased from 8.14% to 8.30% and unsaturated fatty acid (UFA) decreased from 92.12% to 90.88%, but the differences were not significant. In our study, the PUFA/SFA ratio decreased in all varieties and this decrease is thought to be caused by the oxidation of linoleic acid.

During storage, free fatty acid levels increased from 0.09% to 0.45% in Tombul, from 0.11% to 0.40% in Kara and from 0.20% to 0.35% in Sivri (p ˂ 0.001) (Table 3). At the end of the storage period, all varieties were found to have the same level of free fatty acid (p ˃ 0.05). It has been reported that as a result of 18 months’ storage of in-shell Çakıldak hazelnut species, the free fatty acidity increased from 0.09% to 0.48% (Turan and Islam, 2016); whereas in a similar study it was reported to increase from 0.06% to 0.47% (Ghirardello et al., 2013). It has been reported that exceeding 0.7% of free fatty acid concentration in oleic acid in hazelnut indicates the onset of rancidity (Çam and Kılıç, 2008). The results in our work are below these values.

At the beginning of storage, the hazelnuts exhibited 0.00 peroxide values, and increased during storage in all varieties and reached 0.39 for Tombul, 0.30 for Kara and 0.65 for Sivri (p ˂ 0.001) (Table 3). UFA concentration is one of the factors affecting the rate of oil oxidation (Çam and Kılıç, 2008). It is thought that for this reason the Sivri variety, which has the highest amount of MUFA, reached the highest peroxide value at the end of storage. In a study carried out in natural hazelnuts for 18 months, the peroxide values of the samples increased from 0.21% to 2.01% (Koç Güler et al., 2017a). In another study in which shelled hazelnuts were stored for 18 months, it increased from 0.03% to 0.62% (Turan and Islam, 2016). In a study where hazelnuts were stored for 12 months, the initial 0.05% peroxide value increased to 0.26% during storage (Ghirardello et al., 2013). In a study in which the storage stability of hazelnut flour was determined, it was reported that the peroxide value increased during storage (Evren, 2011). It is stated that rancidity is not perceived until reaching peroxide values above 2.0 in hazelnuts (Bostan and Koç Güler, 2016). Findings in our study are well below this value.

The oleic acid/linoleic acid value increased in all varieties at the end of the storage period (p ˂ .001). The highest O/L value at the beginning and end of storage was determined for Sivri (11.59-12.74), Sivri was followed by Tombul (9.66-10.48) and Kara (7.08-7.42) and the variation was statistically significant (p ˂ 0.001) (Table 3). Among the varieties, the highest O/L value in Sivri was due to its highest oleic acid and lowest linoleic acid contents while the lowest value was for Kara because this variety has the lowest oleic acid and the highest linoleic acid amounts. In one study, the O/L value was determined as 13.54-12.78 (Belviso et al., 2017) and in another study it was determined as 9.90 for Tombul, 9.00 for Kara and 11.30 for Sivri (Alasalvar et al., 2010). The oleic/linoleic ratio is an important criterion for assessing kernel quality, and a higher O/L value means better oxidative stability (Alasalvar et al., 2010). The O/L value increased with decreasing linoleic acid content, and the oxidative stability of all varieties increased at the end of storage. Among the varieties, it is thought that Sivri has the highest oxidative stability and storage stability at the beginning and end of storage compared to the others.

Iodine value is a measure of the degree of unsaturation of lipids. High IV is an indication that oils are more reactive, less stable and more susceptible to rancidity and oxidation (Belviso et al., 2017). The highest IV at the beginning of storage was found for Kara (94.40), followed by Tombul (91.61) and Sivri (89.81) (Table 3). At the end of storage, the Tombul’s IV decreased from 91.61 to 90.83 (p ˂ 0.001), but there was no change in the Kara or Sivri (p ˃ 0.05). In one study, it was determined that the IVs of natural hazelnuts were 89.84-86.94 (Belviso et al., 2017).

According to these results, it can be said that at the end of the storage period, the oxidation resistance of Tombul increased and the other varieties it did not change.