Biochemical composition and shape-dimensional traits of rosehip genotypes

Seed-propagated rosehip genotypes in Mesudiye (Ordu) and Talas (Kayseri) districts constituted the material of the present study. Mesudiye has an altitude of 1,135 m and a transitional climate between semi-arid and semi-humid climates. Talas has an altitude of 1,148 m and a dominant terrestrial Central Anatolia climate (Anonymous, 2020). The initial 16 genotypes are located in Mesudiye and 9 genotypes are located in Talas (a total of 25 genotypes were used in this study). From each genotype, 100 fruits were collected, placed into plastic bags and brought to the laboratory in a cooler.

Biochemical analyses were conducted in 5 replicates with 20 fruits in each replicate. The fruits were deseeded with a stainless-steel blade and homogenised in a food blender. Homogenised fruit samples were placed into falcon tubes (about 50 g) and preserved at −20°C until the performance of bioactive analyses.

Fruit DPPH antioxidant activity was determined with use of the modified version of Brand-Williams et al. (1995) method. For analysis, 0.26 mM DPPH (1,1-diphenyl-2-picryl-hydrazyl) solution was prepared. About 100 μL fruit extract was supplemented with 2,900 μL ethyl alcohol and 1 mL DPPH solution, vortex-mixtures and kept in the dark for 30 min. Following incubation, sample absorbance was read in a spectrophotometer at 517 nm wavelength. The resultant absorbance values were expressed in μmol Trolox (10–100 μmol · L⁻¹) equivalent fresh weight (μmol · kg⁻¹).

The total flavonoids in the sample were determined following the method of Chang et al. (2002). About 1,000 μL of fruit extract sample was supplemented with 3.3 mL methanol, then supplemented with 0.1 mL 10% AlCl₃·6H₂O and CH₃COOK. Sample absorbance was read in a spectrophotometer at 415 nm wavelength. Total flavonoids were expressed in quercetin equivalent (QE), mg · kg⁻¹ fresh weight.

Fruit total phenolics was determined with the use of Folin-Ciocalteu reagent. Initially, 500 μL of fresh fruit extract was supplemented with 4.2 mL distilled water, then with 100 μL Folin-Ciocalteu reagent and 2% sodium carbonate (Na₂CO₃). The resultant solution was incubated for 2 h and readings were performed in a spectrophotometer at 760 nm wavelength. Total phenolics was expressed in gallic acid equivalent mg · kg⁻¹ (fresh weight) (Beyhan et al., 2010).

Randomly, 35 samples were taken from each genotype, which was encoded as G1–G25 (Figure 1) to determine the shape and dimensional traits. Samples were placed on a fibreglass plate in a 5 × 7 matrix arrangement and *.tiff extension images were taken using a Nikon D90 model camera. Artificial lighting was provided beneath the plate to prevent shadow formation while imaging (Ercisli et al., 2012). The camera was fixed on a tripod and images were taken from 50 cm above the samples. An external shutter release was used to prevent vibration of the camera. Imaging was conducted at both horizontal and vertical orientation for 3-D dimensional analysis.

Figure 1

The SigmaScan Pro v.5.0 software was used to determine the shape and dimensional properties of the rosehip genotypes. With the image processing analysis, length (L, mm), width (W, mm), thickness (T, mm), projected area (PA, mm²), equivalent diameter (ED, mm), perimeter (P, mm) and circularity (C) values were directly measured. The dimensions and area measures are presented in Figure 2. With the use of L, W and T values, geometric mean diameter (D_g, mm), horizontal elongation (E_h) and vertical elongation (E_v) values were calculated using Eqs (1)–(3), respectively (Mohsenin, 1986; Sayıncı et al., 2015a). (1) Dg=L⋅W⋅T3 {D_g} = \root 3 \of {L \cdot W \cdot T} (2) Eh=LW {E_h} = {L \over W} (3) Ev=WT {E_v} = {W \over T}

Figure 2

Surface area (SA, mm²) and sphericity (φ, %) of rosehip genotypes were calculated as a function of geometric mean diameter using Eqs (4) and (5), respectively (Mohsenin, 1986; Demir et al., 2020). (4) SA=π⋅Dg2 SA = \pi \cdot D_g^2 (5) φ=Dg2L⋅100 \varphi = {{D_g^2} \over L} \cdot 100

The horizontal area measured over 2-D plane is the so-called projected area. Circularity of the genotypes (C) was calculated as a function of projected area (PA, mm²) and perimeter (P, mm) using Eq. (6). A circularity value of 1 indicates a full-circular shape of the material (Sayıncı et al., 2015a). (6) C=4⋅π⋅PAP2 C = 4 \cdot \pi \cdot {{PA} \over {{P^2}}} The volume (V) of geometrically ellipse-like fruits was calculated using the formula for the volume of an ellipse Eq. (7). The ratio of the projected area at horizontal orientation to geometric surface area was defined as the surface closure rate (SCR) and calculated using Eq. (8). When L and W are the same, the SCR equation is defined as projected area/area of circle. Otherwise, when L and W are different, the SCR value is projected area/area of ellipse. An SCR value of 1 indicates that the projected area of the fruit closed the entire surface area calculated based on the largest dimensions (Demir et al., 2019). (7) V=16⋅π⋅L⋅W⋅T V = {1 \over 6} \cdot \pi \cdot L \cdot W \cdot T (8) SCR=4⋅PAπ⋅L⋅W SCR = {{4 \cdot PA} \over {\pi \cdot L \cdot W}}

For elliptical Fourier analysis (EFA), at least 70 images of each genotype were used. Analyses were conducted using the SHAPE (version 1.03) software (Iwata and Ukai, 2002). This analysis comprises definition of contours of a closed shape, identification of the x and y coordinates of the points on the curve constituting a shape, conversion of coordinate values into a mathematical function and identification of function coefficients (Sayıncı, 2016). The function coefficients depend on the number of harmonics and the present analyses were conducted over 20 harmonics. Each harmonic generates four Fourier coefficients (a_n, b_n, c_n and d_n). The a_n and b_n coefficients correspond to the x coordinate and the c_n and d_n coefficients correspond to the y coordinate of the curve (Neto et al., 2006; Ozkan-Koca, 2012).

For image processing, rosehip image files were converted into 24-bit *bmp format. Four modules were used to obtain the shape data. In the Module I (ChainCoder), image processing and shape contour codes were generated. In Module II (Chc2Nef), contours were normalised and elliptic Fourier descriptors were obtained. In Module III (PrinComp), descriptors were subjected to principal component (PC) analysis and PC scores were obtained. In Module IV (PrinPrint), the shape variations of fruit image contours were visualised.

Statistical analyses were conducted using the SPSS 23.0 software. Means for biochemical traits were compared using Duncan's test at a 5% significance level. The shape and dimensional properties of rosehip genotypes were explained with box-plot graphs. On these graphs, extreme values, means and medians were indicated with symbols and mean, standard deviation, minimum and maximum values of each variable were presented. Extreme values were not included in the minimum and maximum values. Differences in shape and dimensional traits of rosehip genotypes were identified with the use of PC analysis. The most significant variables designating the differences in shape and dimensional traits were ordered based on the factor loads. Differences between the genotypes were presented in scatter plots based on component scores. The contour codes obtained through EFA were normalised and multivariate variance analysis (MANOVA) was conducted to test the shape differences in the genotypes. The PAST v.4.02 software was used for MANOVA. The shape differences in the genotypes were explained by Hotelling's paired comparison tests, including verified Bonferroni values and Mahalanobis distances. In linear discriminant analysis conducted with the use of PC scores, the functions revealing shape differences of the genotypes were determined and similarity relations between the genotypes were presented in scatter plots. Such similarities were also put forth by hierarchical clustering analysis with the use of Euclidean similarity index and shape-similar genotypes were grouped on a dendrogram.

Differences in antioxidant activity, total flavonoids and total anthocyanins of seed-propagated rosehip fruits collected from two different locations were found to be significant (p < 0.05) (Table 1).

Antioxidant activity of rosehip genotypes varied from 39.510 mmol · kg⁻¹ (G6) to 72.673 mmol · kg⁻¹ (G19). In terms of antioxidant activity, G19 was respectively followed by G20 (67.944 mmol · kg⁻¹), G25 (67.705 mmol · kg⁻¹) and G24 (64.864 mmol · kg⁻¹). There were significant variations in antioxidant activity of the genotypes and those collected from Kayseri province generally had greater antioxidant activity. In previous studies, rosehip genotypes showed strong DPPH radical (2,2-diphenyl-1-picrylhydrazyl) scavenging activity (Yolcu, 2010). Using the DPPH method, the antioxidant activity values for methanol extracts of samples were reported to be between 79.16% and 87.78% (Fattahi et al., 2012) and between 62.6% and 93.4% (Orhan et al., 2012). The antioxidant capacity of rosehip fruits was also determined through DPPH reducing power of the solution prepared with trolox or ascorbic acid standards. In such studies, the DPPH radical scavenging activity of rosehip fruits was reported to be 278.90 μmol TE · g⁻¹ for methanol extract samples (Demir et al., 2014), respectively, as 32.7 μg TE · mL⁻¹ and 21.7 μg TE · mL⁻¹ for water and methanol extracts (Nadpal et al., 2016), as between 4.83 μmol AAE · g⁻¹ and 5.26 μmol AAE · g⁻¹ (Kasun, 2017) and as between 14.2 μg TE · mL⁻¹ and 31.1 μg TE · mL⁻¹ (Beyhan et al., 2017) for water–methanol (1/1) extracts. Layina-Pathirana et al. (2006) indicated that DPPH free-radical scavenging-based analysis was more advantageous over the other methods in antioxidant activity analysis. On the other hand, different methods have been used to determine the antioxidant activity of rosehip fruits. For the antioxidant capacity of rosehip fruits, Su et al. (2007) used the ABTS⁺ method and reported the values to be between 190 μmol TE · g⁻¹ and 370 μmol TE · g⁻¹; Demir et al. (2014) reported the antioxidant activity to be 35.51 μmol TE · g⁻¹ with ABTS⁺ method and as 301.80 μmol TE · g⁻¹ with FRAP method; Murathan et al. (2016) used the FRAP method and reported the value as 97.75 μmol TE · g⁻¹; Eroglu and Oguz (2018) also used the FRAP method and reported the values to be between 56.80 μmol TE · g⁻¹ and 13.60 μmol TE · g⁻¹. The values of the present study related to DPPH activity were greater than the majority of previous studies and the differences were mainly attributed to the difference in the ecologies in which the plants grow, growing conditions, ripening levels and extraction methods (Wu et al., 2004; Ozturk et al., 2009; Alp et al., 2016).

The greatest total flavonoids were obtained from the genotypes G18 (1,686.20 mg QE · kg⁻¹) and G19 (1,505.20 mg QE · kg⁻¹) collected from Kayseri province. The lowest values were obtained from the genotypes G7 (287.80 mg QE · kg⁻¹) and G11 (292.80 mg QE · kg⁻¹) (Table 1). The present findings revealed quite a large variation in the total flavonoids of rosehip fruits. Similar findings were also reported in previous studies conducted with rosehip fruits. The total flavonoids of rosehip genotypes collected from different parts of Iran were reported as 10.4 mg QE · g⁻¹ (Montazeri et al., 2011); as between 41.0 mg QE · 100 g⁻¹ and 72.0 mg QE · 100 g⁻¹ in Poland (Adamczak et al., 2012); as 196.26 mg rutin · g⁻¹ (Tumbas et al., 2012) and 38.52 mg QE · g⁻¹ (Paunovic et al., 2019) in Serbia; as between 101.3 mg QE · 100 g⁻¹ and 163.2 mg QE · 100 g⁻¹ (Roman et al., 2013) and as between 211.8 mg QE · 100 g⁻¹ and 672.67 mg QE · 100 g⁻¹ (Soare et al., 2015) in Romania; as between 151.0 mg QE · 100 g⁻¹ and 241.0 mg QE · 100 g⁻¹ in Sivas province of Turkey (Beyhan et al., 2017) and as between 29.5 mg QE · 100 g⁻¹ and 36.3 mg QE · 100 g⁻¹ in Samsun province of Turkey (Tastekin, 2017). The differences in total flavonoids of rosehip fruits were mainly attributed to the differences in genotypes, ecological conditions and extraction methods.

Total phenolics of the genotypes varied between 38,519.40 (G11) mg GAE · kg⁻¹ and 79,080.60 (G19) mg GAE · kg⁻¹ with a large variation (Table 1). In gallic acid equivalent fresh weight, the present total phenolics were greater than the findings of Fattahi et al. (2012), who reported the total phenolics in Iran as between 1,764.8 mg and 2,256.5 mg; the findings of Demir et al. (2014) (31,080 mg) and Beyhan et al. (2017) (between 3,400 mg and 4,640 mg) in Turkey were analogous with the findings of Yolcu (2010) (41,846 mg GAE · kg⁻¹), Murathan et al. (2016) (62,980 mg GAE · kg⁻¹) and Tastekin (2017) (68,454 mg GAE · kg⁻¹) in Turkey, Soare et al. (2015) (41,750 mg GAE · kg⁻¹) in Romania and Taneva et al. (2016) (69,000 mg GAE · kg⁻¹) in Bulgaria. On the other hand, Yilmaz and Ercisli (2011) reported the total phenolics of rosehip fruits grown in Turkey as between 78,000 mg GAE · kg⁻¹ and 102,000 mg GAE · kg⁻¹, and Aptin et al. (2013) reported the total phenolics of 30 rosehip genotypes collected from different regions of Iran as between 57,000 mg GAE · kg⁻¹ and 152,000 mg GAE · kg⁻¹. In other studies, conducted on rosehip genotypes, the total phenolics were reported as 99,820 mg GAE · kg⁻¹ in Gümüşhane province of Turkey (Yildiz and Alpaslan, 2012) and as 90,510 mg GAE · kg⁻¹ in Serbia (Paunovic et al., 2019).

The present findings on the antioxidant activity, total flavonoids and total phenolics revealed that there were significant variations between the genotypes and such values were influenced by the province from where they were collected and also the background of the genotypes. Previous studies also indicated that genotypes, altitude, soil and climate conditions, ecological conditions, fruit ripening levels and extraction methods strongly affect fruit contents (Serce et al., 2010; Eroglu and Oguz, 2018).

The general shape and dimensional traits of rosehip genotypes are presented in Figure 3. In terms of dimensional traits, fruit lengths were generally greater than the width and thickness values. The present values revealed that rosehip genotypes had an ellipse-like shape. The present findings on the dimensional traits comply with the findings of Demir and Özcan (2001). Equivalent diameter is calculated based on the projected area. The geometric mean diameter had a lower average than the equivalent diameter. Since the dimensional traits were measured on a 3-D plane, the fruit diameter is the best explained with the geometric mean diameter (Sayıncı et al., 2015b).

Figure 3

The average projected area measured at horizontal orientation was lower than the value measured at vertical orientation. This trait indicated that the rosehip fruits were positioned at a horizontal plane in dimensioning, classification, drying etc. The surface area plays a great role in calculation of the heat transfer rates in drying systems (Bart-Plange et al., 2012). Compared to cherry laurel fruits with an average surface area of 1,230 mm² (Sayıncı et al., 2015a), rosehip fruits had a lower average surface area (674.6 mm²). In this sense, it was thought that the drying duration of rosehip fruits would be shorter compared to cherry laurel fruits. In terms of fruit volume, cherry laurel (4.13 cm³) has 2.5 times greater volume than rosehip fruits (1.66 cm³) (Sayıncı et al., 2015a). The average perimeter of rosehip fruits was calculated as 56.4 mm and such value was quite close to the average perimeter of cornelian cherry fruits (54.3 mm) (Demir et al., 2020).

Greater elongation at horizontal orientation than at vertical orientation revealed that the fruit shape looked like an ellipse. Thus, the circularity and sphericity averages were calculated as 0.712% and 71.4%, respectively. In terms of sphericity, rosehip genotypes were close to cornelian cherry genotypes (78.8%) (Demir et al., 2020). The SCRs varied between 0.83 and 0.98. This ratio may be especially significant in terms of attachment of a fruit onto a perforated surface of pneumatic systems with the aid of air flow. A ratio of 1 indicates that the hole was fully closed by the fruit.

The factor loads for shape and dimensional traits are provided in Table 2. Three PCs were able to explain 98.571% of the total variation. The most important factors differentiating rosehip genotypes were identified as dimensional traits (surface area, geometric mean diameter and volume) gathered under PC1. The factors included in PC2 and PC3 define the shape traits of the genotypes (elongation, sphericity, circularity and SCR). Among these variables, it is remarkable that the elongation factor had negative correlations with PC2.

According to Figure 4A, in terms of surface area, geometric mean diameter and volume, the genotypes G11, G12, G15 and G24 had the greatest averages. The genotypes G3, G8, G9, G13 and G16 had the least averages and were placed on the left of PC1 axis. The greatest sphericity and circularity averages were observed in genotypes G4, G5, G7, G13 and G18. The greatest elongation averages, explaining the ratio of length and width dimensions, were observed in genotypes G3, G11, G16 and G17. According to Figure 4B, the greatest SCRs were observed in genotypes G3, G9, G12, G16, G18 and G21 and the genotypes with the lowest averages were presented in a circle beneath the PC3 axis.

Figure 4

The first three PCs identified based on shape contour codes explained 91.56% of the total variation in shapes of rosehip genotypes (Figure 5). The average shape contour looks like an ellipse. PC1 explained the greatest portion of total variation (82.65%). However, when the ±2 standard deviation of a sample (SD) range was evaluated, it was seen that genotypes had different geometries from each other as of thin/long and sphere. There is a large variation in the transverse shape change (contraction and expansion). PC2 explained 6.38% of the total variation. This variation explained tapering and flattening at the fruit base. PC3 explained 2.53% of the total variation. This component indicated that there was an asymmetric shape change between the genotypes on the horizontal plane. The genotypes constituting this variation had a stoop appearance. These findings play a great role in identification of opening shapes in classification and separation systems (Demir et al., 2020).

Figure 5

The first three functions identified with linear discriminant analysis were able to discriminate 96.7% of shape variations between the genotypes (Table 3a). The first function had the greatest ratio of discrimination (81.8%). The second and third functions had discrimination ratios of 10.6% and 4.3% for shape differences, respectively. According to Table 3b presenting the MANOVA results, shape differences between rosehip genotypes were highly significant (p < 0.001). The pairwise shape differences were analysed with paired Hotelling's test and the results are provided in Table 3c. The verified Bonferroni results given in the bottom triangle revealed that almost all of the genotype pairs had highly significant shape differences. In this test, the shape differences only between G6 and G9–G14 genotypes and G14–G19 genotypes were not found to be significant. The similarities and differences in genotypes pairs could more clearly be seen through the Mahalonabis distances provided in the top triangle of Hotelling's test. Similarity increases as the Mahalonabis distance approaches to 0. It could clearly be seen in terms of the shape that genotype G18 was different from the others.

Figure 6A and 6B presents the scatter plot for discriminant functions, genotypes G13 and G18 were placed on right side of Function 1 axis and the outermost position. It is remarkable that these genotypes had a spherical shape. The genotype G17 was placed on the left side of Function 1 axis and the furthest position, but still beneath the Function 3 axis. In terms of shape, this genotype had an asymmetric appearance on the longitudinal plane. Although G22 genotype was close to the centroid of Functions 1 and 2, it was far from Function 3. This genotype had an ellipse shape.

Figure 6

The shape similarities and differences presented in the scatter plots were proved with hierarchical cluster analysis. As can be seen in Figure 7, the dendrogram had two main groups (I and II). Both groups had three subgroups. The closest genotypes were identified as G14 and G19. This finding complies with the paired comparison tests and scatter plots. In previous studies, clustering analysis was conducted in walnuts (Demir et al., 2018) and cherry laurel (Sayıncı et al., 2015a) and shape differences were successfully put forth.

Figure 7