This study was conducted in 2016 and 2017 on the experimental fields of the Hazelnut Research Institute (Giresun, Türkiye) (40° 54’ 31’’ N and 38° 21’ 09’’ E, 5 m altitude). Plant materials consisted of 22-year-old ‘Tombul’, ‘Palaz’, ‘Çakıldak’, ‘Allahverdi’ and ‘Okay 28’ hazelnut cultivars (Corylus avellana L.). Hazelnut plants were planted in the ‘Ocak’ system at 3-m row spacing and 3-m plant spacing in each row. Cultural practices (fertilization, irrigation, pruning, etc.) were performed at regular intervals. The soil in the experimental area consisted of clay-loam in texture with a pH of 4.78 and organic matter content of 4.58%. Based on soil analysis results, commercial fertilizer containing 20% nitrogen, 22% phosphorus, 15% potassium, 2.0% magnesium oxide, 0.5% zinc and 0.3% boron was applied to each ‘Ocak’ (5 plants in each Ocak) during the second week of February. About 1 kg of fertilizer was applied to each ‘Ocak’. In the second week of May, 0.5 kg of 26% calcium ammonium nitrate fertilizer was applied per ‘Ocak’. Throughout the experiments, pest (powdery mildew, nut weevil and green stink bugs, etc.) and disease control, pruning and weed control were carried out at regular intervals. Figure 1 shows the climate data recorded throughout the experiments.

The experiments were conducted in a randomized-plot experimental design with three replicates for each cultivar and three ‘Ocaks’ (5 plants in each Ocak) in each replicate. Present cultivars were manually harvested at 7 weekly intervals (H1, H2, H3, H4, H5, H6 and H7) from complete kernel growth to complete shell-fill (in all hazelnut cultivars, first harvest was performed July 20 in 2016 and July 24 in 2017). Harvested fruits were manually separated from the husks and sun-dried on a concrete floor until the moisture content dropped to 6%. About 1 kg of whole nuts was used in each replicate from each cultivar for the analyses of each harvest period. The hazelnut shells were cracked and stored at -18 °C until analysis.

Protein content was determined according to the Kjeldahl method and the amount of nitrogen was calculated from the amount of ammonia. The results were expressed in percentage (%) (Venkatachalam and Sathe, 2006).

Oil content was determined according to the Soxhlet extraction method. The results obtained were expressed in percentage (%) (Firestone, 1997).

Fatty acid methyl esters (FAMEs) were prepared for gas chromatography (GC) analysis from the total oil content of hazelnuts using a modified version of the protocol outlined below. First, 1 mL of oil was added to a tube, followed by 2 mL of H₂SO₄ (dissolved in 10% Methanol). After incubating for 40 minutes at 57 °C at 140 rpm, the mixture was cooled to room temperature. Then, 1 mL of 2.0% NaHCO₃ was added and vortexed, 1 mL of hexane was added and the mixture was shaken for one minute. Finally, the FAME-containing upper hexane layer was transferred to a new tube and stored at -20 °C for GC analysis. The Shimadzu GC-20A (Kyoto, Japan) GC with a flame-ionization detector was used to analyze samples filtered through a 0.2 m nylon membrane. For analysis, a Stabilwax DA column (0.25 mm x 0.25 m 60 m) was used. The carrier gas was nitrogen and the flow rate was 3 mL/min. The initial temperature was set at 100 °C, held for four minutes and then raised to 245 °C (20 °C/min) and held there for 40 minutes. After that, the temperature was raised to 250 °C for five minutes. At 250 °C, a split injection (1:20) was performed. Peaks of fatty acids were defined using reference standards by comparing retention times. Results are expressed in a relative percentage of fatty acids and processed using the GC manufacturer’s “GC Solution” program.

Total phenolics, total flavonoids and antioxidant activity (according to DPPH and FRAP) were determined in defatted kernel samples. Oil extraction from the kernel samples was performed using the Soxhlet method. About 1 g defatted kernel samples was weighed on a precise balance (±0.01 g) and 10 mL methanol were added. The prepared solution was kept at +4 °C for two days. The solution was then centrifuged for 4 min at 1200 rpm.

Data normality was checked with the use of Kolmogorov- Smirnov’s test. Variance homogeneity was checked with the use of Levene’s test. Descriptive statistics were calculated. Experimental data were subjected to variance analysis. Significant means were compared with the use of Tukey’s multiple comparison tests (p ≤ 0.05). Statistical analyses were performed with the use of Minitab® 17 Statistical Software (Minitab Inc., State College, PA, USA).

The protein contents should be known in order to effectively time the application of nitrogen fertilizers (Wei and Zhai, 2010). Protein content significantly varied with harvest time (p < 0.05). Protein content fluctuated in all cultivars depending on harvest time. Except for ‘Okay 28’ (16.68%), in other cultivars, the highest protein content was determined in the H1 harvest time (16.71% in ‘Tombul,’ 16.88% in ‘Palaz,’ 18.66% in ‘Çakıldak,’ and 17.60% in ‘Allahverdi’). It was also detected in the H5 in ‘Okay 28’. Protein content decreased in general at the final harvest time (H7) as compared to the first harvest time (H1) (Table 1). Ilyasoglu (2015) and Seyhan et al. (2007) reported that the protein contents in ‘Tombul’, ‘Palaz’, ‘Badem’ and ‘Sivri’ cultivars decreased as the harvest time progressed. Cristofori et al. (2015), on the other hand, stated that the protein content in the ‘Tonda Gentile Romana’ cultivar fluctuated as the harvest time progressed. A decrease was also reported in ‘Tonda di Giffoni’ and ‘Nocchione’ cultivars. The present findings on protein ratio comply with the findings of previous studies.

The period in which hazelnut kernels have maximum oil levels is critical for detecting the optimal harvest time (Bouali et al., 2013). Harvest time had a significant effect on the oil contents in the investigated hazelnut cultivars (p < 0.05). All cultivars had fluctuating oil contents with harvest times. The highest oil contents were determined in the H7 in ‘Tombul’ (68.66%), in the H4 in ‘Çakıldak’ (66.40%); in the H5 in ‘Palaz’ (68.65%), ‘Okay 28’ (69.86%) and ‘Allahverdi’ (65.78%) (Table 1). Although oil contents fluctuated with harvest times, they generally increased as maturation progressed. Reserve lipid formation at the final stage of fruit maturation resulted in high oil contents. However, due to lipids being synthesized during the early stages of kernel development in nut species and being used to form new fruit tissues, oil accumulation during the first stage of kernel development may be lower (Bouali et al., 2013). Consistent with the present findings, it was reported that oil level increased during the kernel development of the cultivars ‘Tombul’, ‘Palaz’ and ‘Sivri’ cultivars (Ilyasoglu, 2015). Similar findings were reported for ‘Tombul’, ‘Palaz’ and ‘Badem’ cultivars (Seyhan et al., 2007), ‘Tonda Gentile Romana’, ‘Tonda di Giffoni’ and ‘Nocchione’ cultivars (Cristofori et al., 2015).

Hazelnut oil is a good source of food which is enjoyed by consumers. Oleic acid is the main fatty acid in hazelnut oil, followed by linoleic, palmitic and stearic acids. These fatty acids account for 98-99% of the total fatty acids in hazelnuts, with other fatty acids found in trace amounts (Balta et al., 2006; Karaosmanoglu and Ustun, 2021; Karakaya et al., 2023). Human diets should have a high oleic acid content and low-density lipoprotein lowers cholesterol levels (Wani et al., 2020). Several factors including variety, ecological conditions, cultural practices, and harvest time affect the fatty acids in hazelnuts (Balta et al., 2006; Cristofori et al., 2015; Balık, 2021).

The effects of harvest time on the oleic acid content in hazelnut cultivars were found to be significant (p < 0.05). The oleic acid contents in the cultivars fluctuated (in the form of an increase-decrease-increase) with harvest time. The highest oleic acid contents were determined in the H3 in the ‘Tombul’ (79.75%), ‘Palaz’ (81.87%), ‘Çakıldak’ (82.13%) and ‘Okay 28’ (82.63%) cultivars. However, it was also detected in the H4 in ‘Allahverdi’ (80.03%). The oleic acid contents in the ‘Tombul’ and ‘Allahverdi’ cultivars increased at the last harvest time (H7) as compared to the first harvest time (H1). In contrast, it decreased in ‘Palaz’ and ‘Okay 28’ cultivars (Table 2). The stearic acid contents in ‘Palaz’ and ‘Okay 28’ cultivars increased, while their oleic acid contents decreased with the progress of harvest time. This can be explained by the conversion of oleic acid to stearic acid by Δ9-stearoyl-ACP from desaturase enzymes (Salas et al., 2000). Again, the decrease in oleic acid content in these cultivars may be related to oleic acid conversion into linoleic acid. Temperature has a direct effect on the activity of the 9-stearoyl-ACP desaturase enzyme, which converts oleic acid into linoleic acid (Bouali et al., 2013). In fact, higher linoleic acid content was determined in these cultivars at the last harvest time (H7) as compared to the first harvest time (H1). Similarly, Cristofori et al. (2015) reported that the effect of harvest time on oleic acid in hazelnut varies by variety. They noted an increase-decrease in oleic acid content in ‘Tonda Gentile Romana’ cultivar as the harvest progressed, as well as a decrease-increase in ‘Tonda di Giffoni’ and ‘Nocchione’ cultivars. Ciemniewska-Zytkiewicz et al. (2015) stated similar results for oleic acid content in the ‘Katalonski’ hazelnut cultivar. In contrast to present findings, Seyhan et al. (2007) and Ilyasoglu (2016) reported increasing oleic acid contents for ‘Tombul’, ‘Palaz’, ‘Badem’ and ‘Sivri’ cultivars as harvest time progressed.

The linoleic acid contents in the investigated hazelnut cultivars varied with harvest time (p < 0.05). The highest linoleic acid was determined in the H4 in ‘Tombul’ (10.95%) and ‘Palaz’ (11.12%); in the H2 in ‘Çakıldak’ (11.63%) and ‘Okay 28’ (8.72%); in the H1 in ‘Allahverdi’ (16.54%). Except for ‘Allahverdi’, linoleic acid content was higher in other cultivars at the last harvest time (H7) as compared to the first harvest time (H1) (Table 2). Different researchers reported that linoleic acid content decreased-increased in ‘Tonda Gentile Romana’ and ‘Palaz’ cultivars (Seyhan et al., 2007; Cristofori et al., 2015) as the harvest progressed; decreased in ‘Tombul’, ‘Badem’, ‘Sivri’ and ‘Katalonski’ cultivars (Seyhan et al., 2007; Ciemniewska-Zytkiewicz et al., 2015; Ilyasoglu, 2016); and increased in ‘Tonda di Giffoni’ and ‘Nocchione’ cultivars (Cristofori et al., 2015). The linoleic acid content in the present cultivars was similar to that reported for ‘Tonda Gentile Romana’ and ‘Palaz’ cultivars (Seyhan et al., 2007; Cristofori et al., 2015). However, it was stated that the effect of harvest time on linoleic acid content in hazelnuts may vary depending on the cultivar (Seyhan et al., 2007; Cristofori et al., 2015; Ilyasoglu, 2016).

While the effect of harvest time on palmitic acid was significant in ‘Tombul’ and ‘Okay 28’ cultivars, it was insignificant in the other cultivars (p < 0.05). Palmitic acid contents fluctuated in ‘Tombul’ and ‘Okay 28’ cultivars with harvest time. The highest palmitic acid contents were determined in the H3 in ‘Tombul’ (7.75%); and in the H2 in ‘Okay 28’ (7.82%). It increased in ‘Tombul’ at the last harvest (H7) as compared to the first harvest (H1); while it decreased in ‘Okay 28’ (Table 3). Such a decrease in ‘Okay 28’ cultivar can be explained by the fact that palmitic acid was the primary product of the fatty acid synthesis pathway and thus other fatty acids in the kernels were derived from palmitic acid (Bouali et al., 2013). Furthermore, different researchers reported that the palmitic acid contents in ‘Tombul’, ‘Palaz’, ‘Badem’, ‘Sivri’, ‘Tonda Gentile Romana’ and ‘Tonda di Giffoni’ cultivars decreased as harvest time progressed (Seyhan et al., 2007; Cristofori et al., 2015; Ilyasoglu, 2016). In ‘Nocchione’ and ‘Katalonski’ cultivars, on the other hand, it showed a decrease-increase (Cristofori et al., 2015; Ciemniewska-Zytkiewicz et al., 2015). In the current study, similar to results reported for ‘Nocchione’ and ‘Katalonski’ cultivars, palmitic acid content fluctuated (increase-decrease or decrease-increase) depending on cultivar as harvest time progressed.

The effects of harvest time on the stearic acid content in the investigated hazelnut cultivars were found to be significant (p < 0.05). With the exception of the ‘Çakıldak’ cultivar, the stearic acid content in the other cultivars fluctuated with harvest time. It increased up to the H5 harvest time in the ‘Çakıldak’ cultivar, then decreased. The highest stearic acid contents were seen in H5 in ‘Tombul’ (4.91%), ‘Çakıldak’ (4.21%) and ‘Allahverdi’ (4.46%); in H7 in ‘Palaz’ (4.30%); in H4 in ‘Okay 28’ (4.72%). Except for ‘Çakıldak’ cultivar, the other cultivars had significantly higher stearic acid contents at the last harvest (H7) than at the first harvest (H1) (Table 3). The stearic acid contents in the different hazelnut varieties were reported to vary depending on the variety as harvest progressed (Seyhan et al., 2007; Cristofori et al., 2015; Ilyasoglu, 2016). Stearic acid contents were determined to increase as harvest time progressed in the ‘Tombul’, ‘Palaz’ and ‘Badem’ cultivars (Seyhan et al., 2007); while they decreased in ‘Tonda Gentile Romana’ and ‘Tonda di Giffoni’ cultivars. It was also determined that stearic acid contents decreased-increased in ‘Nocchione’ (Cristofori et al., 2015) and ‘Katalonski’ cultivars (Ciemniewska-Zytkiewicz et al., 2015) and increased-decreased in the ‘Palaz’ cultivar (Ilyasoglu, 2016).

Hazelnut phenolics protect kernels from oxidation and affect flavor formation in fresh hazelnuts (Delgado et al., 2010). Many researchers have reported that phenolic compounds with antioxidant properties may have a positive effect on human health (Karaosmanoglu and Ustun, 2021). Hazelnuts have high antioxidant capacity (Chang et al., 2016). Therefore, consumers prefer hazelnuts with high phenolic compound levels. However, phenolic compounds in hazelnut can be influenced by a variety of factors, including cultivar, environmental conditions, cultural practices and harvest time (Pycia et al., 2020; Balık, 2021; Karakaya et al., 2023).

The total phenolics in the present hazelnut cultivars were significantly affected by harvest time (p < 0.05). Except for ‘Okay 28’ and ‘Allahverdi’ cultivars, the total phenolics in the other cultivars fluctuated with harvest time and generally decreased after the H3 stage. They decreased up to H5 in the ‘Okay 28’ cultivar and then fluctuated. They increased up to H3 in the ‘Allahverdi’ cultivar, then decreased. The highest total phenolics were seen in H2 in ‘Tombul’ (14.74 g GAE·kg^-1); in H3 in ‘Palaz’ (15.54 g GAE·kg^-1), ‘Çakıldak’ (20.62 g GAE·kg^-1) and ‘Allahverdi’ (18.38 g GAE·kg^-1); in H1 in ‘Okay 28’ (21.28 g GAE·kg^-1) (Table 4).

Changes in total flavonoids with harvest time were significant for all cultivars (p < 0.05). Except for the ‘Tombul’ and ‘Okay 28’ cultivars, the total flavonoids in the other cultivars fluctuated with harvest time. In the ‘Tombul’ cultivar, it decreased after H2. It decreased up to H6 in the ‘Okay 28’ cultivar. The highest total flavonoid contents were determined for the H2 in ‘Tombul’ (3.56 g QE·kg^-1) and ‘Palaz’ (2.85 g QE·kg^-1), and for H1 in ‘Çakıldak’ (3.24 g QE·kg^-1), ‘Okay 28’ (4.91 g QE·kg^-1) and ‘Allahverdi’ (3.38 g QE·kg^-1). Total flavonoids were generally higher at the first harvest (H1) than at the last harvest (H7), except for the ‘Çakıldak’ cultivar (Table 4).

Overall, total phenolics in the cultivars significantly decreased at the last harvest (H7) as compared to the first harvest (H1). This has been associated with an increase in polyphenol oxidase activity during fruit ripening (Parr and Bolwell, 2000). Again, the decrease in phenolic contents in fruits as harvest time progresses is explained by the condensation of different phenolic acids during these periods, followed by the formation of complex phenolic compounds such as tannins and lignin (Ben-Ahmed et al., 2009). Furthermore, ripe fruits have lower total phenolic contents than semi-ripe fruits (Yang et al., 2011). Indeed, many researchers reported the highest total phenolic contents in early-harvested walnut fruit (Wei et al., 2022). Cristofori et al. (2015) stated that total phenolic contents fluctuated (decrease-increase) in ‘Tonda Gentile Romana’ and increased in ‘Tonda di Giffoni’ and ‘Nocchione’, depending on the harvest time. In terms of total phenolics, the present findings were compatible with those obtained from the ‘Tonda Gentile Romana’ cultivar. Similarly, total phenolics were observed to decrease as harvest progressed in walnuts (Pycia et al., 2019) and pistachios (Kelebek et al., 2020). However, some studies showed that the polyphenol concentration in hazelnuts increased as they ripened (Persic et al., 2018). Furthermore, different researchers reported that climate conditions, variety, harvest time, biotic and abiotic stress factors may influence total the phenolics in hazelnuts (Cristofori et al., 2015; Pycia et al., 2020; Karakaya et al., 2023).

High antioxidant capacity of fruits is mostly linked to health benefits. Chang et al. (2016) reported that antioxidant capacity is related to the fruits’ total phenolic contents. The antioxidant activity (both DPPH and FRAP) of the investigated hazelnut cultivars was significantly affected by harvest time (p < 0.05). According to the DPPH test, while the antioxidant activity fluctuated with the harvest time in ‘Palaz’, it decreased in ‘Çakıldak’ and ‘Okay 28’ from the first harvest time (H1). It decreased after H2 in ‘Tombul’. It increased up to H3 in ‘Allahverdi’ and then decreased. The highest antioxidant activities were found in H2 in ‘Tombul’ (34.18 mmol TE·kg^-1), in H1 in ‘Çakıldak’ (35.34 mmol TE·kg^-1) and ‘Okay 28’ (35.13 mmol TE·kg^-1), in H3 in ‘Palaz’ (33.52 mmol TE·kg^-1) and ‘Allahverdi’ (31.07 mmol TE·kg^-1) (Table 5). According to the FRAP test, antioxidant activity fluctuated depending on harvest time in ‘Tombul’, ‘Çakıldak’ and ‘Allahverdi’, while it decreased in ‘Palaz’ and ‘Okay 28’. The highest antioxidant activities were determined in H2 in ‘Tombul’ (60.47 mmol TE·kg^-1); in H1 in ‘Palaz’ (52.45 mmol TE·kg^-1) and ‘Okay 28’ (77.16 mmol TE·kg^-1), in H3 in ‘Çakıldak’ (72.71 mmol TE·kg^-1) and ‘Allahverdi’ (55.14 mmol TE·kg^-1) (Table 5).

According to both methods, the antioxidant activity of the cultivars decreased significantly at the last harvest time (H7) compared to the first harvest time (H1). Similarly, many researchers reported that the antioxidant activity of hazelnuts and walnuts decreased as harvest time progressed (Pycia et al., 2020; Wei et al., 2022). In the current study, antioxidant activity decreased as the harvest time progressed due to a decrease in total phenolics and flavonoids. Indeed, a decrease in antioxidant activity during fruit ripening may be related to a decrease in total phenolic compounds (Pycia et al., 2020).

According to PCA results, the correlation between the investigated traits was 69.1% (PC1 + PC2). Tombul, Palaz and Okay 28 cultivars were associated with oil and stearic acid and were located in the first region in the PCA plane with the H5 harvest time. The Çakıldak cultivar was related to protein and was grouped in the fourth region with the H1 harvest time. The Allahverdi cultivar was associated with linoleic acid and was located in the third region with the H4, H6, and H7 harvest times. Furthermore, bioactive compounds were clustered at the same point in the PCA plane. Protein, oil and fatty acids were also located at different points (Figure 2).