Analysis of Mineral Composition and Isotope Ratio as Part of Chemical Profiles of Apples for Their Authentication

Abstract

One of the consequences of the globalisation of food markets is the effort enabling the control of food security and its origin. This might be traced by using different chemical composition analyses. However, for Central Europe, there is a lack of knowledge about the original reference values as well as their heterogeneity among the lands and countries. This study focused on characterizing the mineral profiles of apple tree fruits and comparing these profiles among different districts in Central Europe. The fruits of the apple cultivars ‘Gala’ and ‘Golden Delicious’ originated in the Czech Republic and Poland. The mineral and isotopic content of the apple fruit flesh was analysed using ICP-MS. The data were processed using the ANOVA test and compositely analysed using the PCA and LDA models. The results show relatively high variation in element distribution, particularly ⁸⁷Sr/⁸⁶Sr, Mn, Zn, Cu, Ca, P, and B, ranging between 20.6 and67.9% for both cultivars on average. However, their high variability within particular districts complicates the resolution of the LDA model. The reasons are linked to the geomorphological and pedological heterogeneity of the analysed districts as well as the particular sensitivity of the set of chosen primers to agronomic practices and tree performance. For this region, only partial separation among districts could be obtained by P, Ca, and Cu content, as well as the isotopic ratio of ¹⁰B/¹¹B. However, the resolution of the geographical discrimination needs to be improved by an enhanced set of primers, the use of more precise analytical techniques for the Sr isotopic ratio, or by multiple chemical analyses. Furthermore, the heterogeneity of the analysed districts could be tackled by more detailed analyses at the level of micro-regions.

1. Introduction

Globalization improves the trade and transfer of production worldwide. Fruits are a part of the food in markets that can be gathered from foreign countries at various distances. On the other hand, this compromises the sustainability of local fruit production, the quality of products, as well as increases their carbon footprint. In line with these global trends, consumers are increasing their preference for local, regional products and their safety [1]. However, to following the geographical authenticity of fruit production, the development, standardization, and implementation of diagnostic procedures needs to be conducted.

Apples are a very important commercial fruit crop worldwide. Due to their relatively undemanding nature, unlike other fruit species, they can be grown in most temperate climates and subtropics with intensive agricultural production. Apples are a key part of a nutritionally balanced diet, especially under the conditions of a temperate climate zone. From the perspective of human diet, they provide delicious fruits that are rich in sugars, minerals, and vitamins [2]. In addition to direct consumption, the fruits are widely used in the food processing industry. The length of maturity for consumption is very long for most of the main cultivars of apple trees, and with the use of modern storage methods, it is possible to supply consumers with apples throughout the year. China is the largest producer of apples, with a roughly fifty percent share of the total annual amount of 95.84 million tons of world apple production for 2022. The USA, Turkey, and EU countries [3] follow at a great distance.

Apple (Malus), like other representatives of pome and stone fruits, belongs to the rose family (Rosaceae Juss.). Individual genotypes of the cultivated apple tree (Malus domestica Borkh.) differ considerably in their growth and bearing habits, as well as their mineral accumulation [4] and fruit chemical content [5]. This is due to the genetic heterogeneity of this fruit species. Results of apple genomic studies have confirmed that cultivated apple varieties originate mainly from Malus sieversii M. Roem found in Central Asia. Other species, such as the European apple (Malus sylvestris Mill.), the small-fruited Malus baccata (L.) Borkh. from Northeast Asia, and the Eastern Caucasian apple Malus orientalis Uglitz., also contributed to the creation of today’s cultivated varieties [6]. From a pomological point of view, apple trees are classified together with pears as pome fruits. Seeds, as generative organs, are stored in the middle of the fruit in a core (endocarp) surrounded by the edible part of the fruit—the pulp covered by skin (exocarp).

Currently, there is a spectrum of methods aimed at determining the authenticity of food, such as DNA-based techniques, physicochemical, isotope, and elemental analyses, spectroscopy, and separation techniques [7]. Their use is primarily related to the purpose of analysis, i.e., determining, for example, the geographical origin of food, verifying the correct composition of the product, or identifying the use of additives. The type of food, the method of obtaining the primary raw material, and its subsequent processing are also important in this context [8,9,10,11]. The analysis of elemental composition, including isotopic ratios, using ICP-MS (Inductively Coupled Plasma—Mass Spectrometry) represents one of the effective methods for determining the geographical authenticity of foods [8,12]. These analytical procedures are based on the different mineral composition of the elements in the soil. The different mineral content is related to different soil-forming processes, environmental conditions, and soil management and fertilization [13,14]. Despite the fact that fruit trees receive mineral elements from the soil based on their current needs [15], their availability in the soil is, in many cases, determined preferentially by the mineral composition of soils [16], which can vary locally [17]. The cultivated crop, its variety, or the rootstock combination used can also play an important role [18,19]. This method has been used in the past to verify the authenticity and quality of fruits [10,11] and fruit juices [9,20,21]. Among the countries of Central Europe, there is increasing demand for authentication of fruit origin because of consumer protection. However, the geographical authenticity assessment method for determining the origin of apples under the conditions of Central Europe, using mineral composition and isotopic ratio analysis, has not yet been investigated in detail. This work presents the results of ICP-MS achieved at the Research and Breeding Institute of Pomology Holovousy Ltd. aimed at determining geographic authenticity through analysis of the mineral content of elements and selected isotopic ratios in apples from the Czech Republic and Poland, which is currently the most important foreign supplier of this fruit commodity to the Czech domestic market. Moreover, the Czech Republic and Poland share a long border, and there are partially similar geological and soil formations, which have similar mineral composition.

2. Materials and Methods

2.1. Plant Material

The fruit samples were collected in the main apple production regions of the Czech Republic and Poland in the following districts: Central Bohemia (CR-CeB), Eastern Bohemia (CR-EaB), South Moravia (CR-SoM), Lower Silesian Voivodeship (PL-LSV), Łódź Voivodeship (PL-LoV), and Masovian Voivodeship (PL-MaV). From these districts, 64 samples were collected in the years 2020–2022. The two main commercial apple cultivars, ‘Gala’ and ‘Golden Delicious’, were selected for the experiment. Samples were collected from the standard commercial distribution channels of particular apple-growing farms.

2.2. Determination of Mineral Content in Fruits

The nutritional status and isotope ratios of the apple trees were determined from the fruits. Each sample consisted of 5–10 fruits collected from the Czech Republic and Poland (Table 1 and Table 2). In the mineral analysis of fruit dry matter, the content of phosphorus (P), potassium (K), magnesium (Mg), calcium (Ca), boron (B), zinc (Zn), manganese (Mn), and iron (Fe) was assessed. In terms of isotope ratios, B as ¹⁰B/¹¹B and Sr as ⁸⁷Sr/⁸⁶Sr were evaluated. The samples were washed in demineralized water, and after evaporation of the droplets of water, they were dried at 50 °C to constant weight in a laboratory dryer. The fruit samples were then ground using Grindomix GM 200 to a powder. The content of mineral nutrients and isotope ratios was determined using ICP-MS (Agilent 7900, Agilent Technologies Inc., USA) after previous nitric acid digestion (microwave-digestion system, Discover SP-D 80, CEM). The sample (0.25 g) was mixed with 6 mL of 65% nitric acid for trace analysis and digested at 200 °C for 4 min. The method was specified by CEM Corporation and verified by the Central Institute for Supervising and Testing in Agriculture, Czech Republic, under number JPP 40033.1 [22]. All reagents and standards used were determined for ICP-MS analysis and fulfilled the quality norms ISO/IEC 17025 and ISO 17034. The analysis of nutritional status and isotope ratio was measured in the laboratory and fulfilled ISO/IEC 17025. The method was validated by the use of the standard reference material NIST 1547.

2.3. Statistics

The obtained profiles of the elemental composition of individual samples were statistically processed using Principal Component Analysis (PCA). This analysis enables a comprehensive evaluation of the mutual relations of the evaluated variables using the so-called components. The closeness of the relationship of individual variables to the calculated components determines their relative importance in the analysis and interpretation of sample profiles. The subsequent analysis of the main components enables the spatial separation of the measured groups of samples and their mutual comparison. The statistical difference among the analysed districts for particular elements or isotopic ratios was tested by analysis of variance (ANOVA). The combination of both methods allowed mapping and evaluating the variables’ significance for the discrimination of samples among the districts. Consequently, linear discriminant analysis (LDA) was used to test the classification of the samples among different districts according to their geographical origin and to compare the reliability of different models according to sets of variables.

Prior to statistical processing, the data were pre-analysed and adjusted to meet the assumptions for the mentioned tests. For PCA, the data were standardized by the centre-then-scale method. The samples were clustered into districts according to their geographical origin. The normality of residuals and homogeneity of variance were tested prior to ANOVA using the exact Shapiro–Wilk and Cochran–Hartley–Bartlett tests, respectively. Results were considered statistically significant at ρ < 0.05. The variability in the elements’ content was further described by the coefficient of variation among all samples, as well as among different districts. Because of unequal variance and non-normal distribution of some of the analysed mineral elements and isotopic ratios, normalization and logarithmic transformation were done prior to ANOVA. The results of ANOVA were then back-transformed using exponential transformation and further compared among the districts using the estimated marginal means test through the “emmeans” library [23]. Statistical significance of the factor district for particular variables was described by the following symbols: “*” at ρ < 0.05, “**” at ρ < 0.01, and for the summary tables using confidential intervals at γ = 95%.

Table 1. Means, confidence intervals (CIs), and coefficients of variation (CVs) of the analysed mineral elements and isotopic ratios in the apple fruit flesh of ‘Gala’ in 2020–2022.

Locality

P (±CI)

CV

K (±CI)

CV

Mg (±CI)

CV

Ca (±CI)

CV

Fe (±CI)

CV

Mn (±CI)

CV

CR-CeB

698 (558–837)

9.67

7944 (6863–9024)

6.60

364 (308–419)

7.36

449 (341–558)

11.71

8.29 (6.59–10.00)

9.99

3.97 (2.05–5.90)

23.48

CR-EaB

677 (594–760)

5.92

7743 (7098–8388)

4.04

346 (314–379)

4.54

460 (392–528)

7.15

7.22 (6.31–8.13)

6.11

2.68 (1.89–3.48)

14.37

CR-SoM

607 (486–728)

9.67

7522 (6499–8546)

6.59

372 (316–429)

7.39

284 (215–352)

11.69

6.74 (5.35–8.12)

9.99

2.49 (1.29–3.70)

23.49

PL-LoV

570 (482–658)

7.49

8187 (7325–9050)

5.11

335 (295–374)

5.70

351 (286–417)

9.09

6.99 (5.87–8.10)

7.73

2.86 (1.79–3.94)

18.18

PL-LSV

529 (460–598)

6.33

7167 (6528–7805)

4.31

308 (277–338)

4.84

361 (304–418)

7.65

6.25 (5.41–7.09)

6.54

2.80 (1.91–3.68)

15.32

PL-MaV

680 (562–797)

8.37

9020 (7957–10,082)

5.71

357 (310–404)

6.39

351 (278–425)

10.14

7.11 (5.84–8.38)

8.65

3.19 (1.85–4.52)

20.31

Locality

Zn (±CI)

CV

B (±CI)

CV

Cu (±CI)

CV

10B/11B (±CI)

CV

87Sr/86Sr (±CI)

CV

CR-CeB

2.49 (1.62–3.36)

16.95

15.7 (12.1–19.2)

11.02

2.83 (1.99–3.66)

14.31

0.193 (0.178–0.208)

3.75

9.87 (3.47–16.3)

31.41

CR-EaB

1.66 (1.30–2.01)

10.36

20.1 (17.3–22.9)

6.77

1.99 (1.63–2.35)

8.79

0.208 (0.198–0.218)

2.29

10.76 (6.49–15.00)

19.24

CR-SoM

1.18 (0.77–1.59)

16.95

17.5 (13.5–21.4)

11.03

1.99 (1.40–2.58)

14.32

0.201 (0.185–0.216)

3.74

15.45 (5.44–25.50)

31.39

PL-LoV

1.75 (1.28–2.23)

13.20

15.3 (12.6–18.0)

8.56

1.33 (1.03–1.64)

11.13

0.186 (0.175–0.198)

2.90

12.47 (6.21–18.70)

24.30

PL-LSV

1.63 (1.26–2.01)

11.17

19.2 (16.3–22.1)

7.24

2.02 (1.63–2.42)

9.41

0.200 (0.190–0.210)

2.45

16.98 (9.77–24.20)

20.55

PL-MaV

1.94 (1.36–2.53)

14.74

21.6 (17.3–25.8)

9.58

2.31 (1.72–2.90)

12.38

0.216 (0.202–0.230)

3.24

24.21 (10.62–37.8)

27.18

Table 1. Means, confidence intervals (CIs), and coefficients of variation (CVs) of the analysed mineral elements and isotopic ratios in the apple fruit flesh of ‘Gala’ in 2020–2022.

Locality

P (±CI)

CV

K (±CI)

CV

Mg (±CI)

CV

Ca (±CI)

CV

Fe (±CI)

CV

Mn (±CI)

CV

CR-CeB

698 (558–837)

9.67

7944 (6863–9024)

6.60

364 (308–419)

7.36

449 (341–558)

11.71

8.29 (6.59–10.00)

9.99

3.97 (2.05–5.90)

23.48

CR-EaB

677 (594–760)

5.92

7743 (7098–8388)

4.04

346 (314–379)

4.54

460 (392–528)

7.15

7.22 (6.31–8.13)

6.11

2.68 (1.89–3.48)

14.37

CR-SoM

607 (486–728)

9.67

7522 (6499–8546)

6.59

372 (316–429)

7.39

284 (215–352)

11.69

6.74 (5.35–8.12)

9.99

2.49 (1.29–3.70)

23.49

PL-LoV

570 (482–658)

7.49

8187 (7325–9050)

5.11

335 (295–374)

5.70

351 (286–417)

9.09

6.99 (5.87–8.10)

7.73

2.86 (1.79–3.94)

18.18

PL-LSV

529 (460–598)

6.33

7167 (6528–7805)

4.31

308 (277–338)

4.84

361 (304–418)

7.65

6.25 (5.41–7.09)

6.54

2.80 (1.91–3.68)

15.32

PL-MaV

680 (562–797)

8.37

9020 (7957–10,082)

5.71

357 (310–404)

6.39

351 (278–425)

10.14

7.11 (5.84–8.38)

8.65

3.19 (1.85–4.52)

20.31

Locality

Zn (±CI)

CV

B (±CI)

CV

Cu (±CI)

CV

10B/11B (±CI)

CV

87Sr/86Sr (±CI)

CV

CR-CeB

2.49 (1.62–3.36)

16.95

15.7 (12.1–19.2)

11.02

2.83 (1.99–3.66)

14.31

0.193 (0.178–0.208)

3.75

9.87 (3.47–16.3)

31.41

CR-EaB

1.66 (1.30–2.01)

10.36

20.1 (17.3–22.9)

6.77

1.99 (1.63–2.35)

8.79

0.208 (0.198–0.218)

2.29

10.76 (6.49–15.00)

19.24

CR-SoM

1.18 (0.77–1.59)

16.95

17.5 (13.5–21.4)

11.03

1.99 (1.40–2.58)

14.32

0.201 (0.185–0.216)

3.74

15.45 (5.44–25.50)

31.39

PL-LoV

1.75 (1.28–2.23)

13.20

15.3 (12.6–18.0)

8.56

1.33 (1.03–1.64)

11.13

0.186 (0.175–0.198)

2.90

12.47 (6.21–18.70)

24.30

PL-LSV

1.63 (1.26–2.01)

11.17

19.2 (16.3–22.1)

7.24

2.02 (1.63–2.42)

9.41

0.200 (0.190–0.210)

2.45

16.98 (9.77–24.20)

20.55

PL-MaV

1.94 (1.36–2.53)

14.74

21.6 (17.3–25.8)

9.58

2.31 (1.72–2.90)

12.38

0.216 (0.202–0.230)

3.24

24.21 (10.62–37.8)

27.18

Table 2. Means, confidence intervals (CIs), and coefficients of variation (CVs) of the analysed mineral elements and isotopic ratios in the apple fruit flesh of ‘Golden Delicious’ in 2020–2022.

Locality

P (±CI)

CV

K (±CI)

CV

Mg (±CI)

CV

Ca (±CI)

CV

Fe (±CI)

CV

Mn (±CI)

CV

CR-CeB

703 (513–893)

13.19

8485 (6889–10,080)

9.18

338 (286–389)

7.40

261 (214–308)

8.77

7.75 (5.99–9.52)

11.11

4.35 (2.51–6.19)

20.64

CR-EaB

668 (574–762)

6.89

8009 (7222–8796)

4.79

310 (285–334)

3.87

260 (235–284)

4.58

6.98 (6.15–7.80)

5.80

2.89 (2.25–3.53)

10.80

CR-SoM

605 (478–731)

10.21

6808 (5816–7800)

7.11

319 (281–356)

5.74

205 (176–233)

6.78

6.51 (5.37–7.66)

8.60

3.47 (2.33–4.61)

15.99

PL-LoV

467 (369–564)

10.21

7443 (6359–8527)

7.11

279 (247–312)

5.73

244 (210–278)

6.80

6.45 (5.31–7.59)

8.60

2.64 (1.78–3.51)

16.02

PL-LSV

457 (362–553)

10.22

7254 (6198–8311)

7.11

290 (256–324)

5.72

260 (224–296)

6.81

6.28 (5.18–7.39)

8.61

3.78 (2.54–5.02)

16.01

PL-MaV

574 (454–695)

10.23

6824 (5830–7819)

7.11

315 (278–352)

5.71

274 (235–312)

6.79

6.88 (5.67–8.10)

8.60

3.60 (2.42–4.78)

16.00

Locality

Zn (±CI)

CV

B (±CI)

CV

Cu (±CI)

CV

10B/11B (±CI)

CV

87Sr/86Sr (±CI)

CV

CR-CeB

2.57 (1.38–3.76)

22.57

17.4 (10.5–24.4)

19.48

3.00 (1.89–4.11)

18.07

0.193 (0.181–0.206)

3.26

21.10 (6.29–36.00)

34.36

CR-EaB

1.58 (1.20–1.96)

11.77

22.0 (17.4–26.6)

10.18

2.16 (1.74–2.58)

9.44

0.208 (0.201–0.215)

1.70

13.10 (8.31–17.90)

17.94

CR-SoM

1.30 (0.83–1.76)

17.46

21.7 (15.0–28.4)

15.12

1.90 (1.36–2.45)

14.05

0.200 (0.189–0.210)

2.52

23.20 (10.59–35.90)

26.59

PL-LoV

1.53 (0.98–2.08)

17.52

15.9 (11.0–20.8)

15.09

1.17 (0.83–1.50)

14.02

0.186 (0.176–0.195)

2.52

11.70 (5.34–18.10)

26.58

PL-LSV

1.88 (1.21–2.55)

17.45

19.2 (13.3–25.1)

15.05

1.78 (1.27–2.29)

13.99

0.191 (0.182–0.201)

2.52

27.00 (12.31–41.70)

26.56

PL-MaV

1.52 (0.98–2.07)

17.50

20.0 (13.8–26.1)

15.05

1.93 (1.38–2.49)

14.04

0.213 (0.202–0.224)

2.52

18.70 (8.54–28.90)

26.63

Table 2. Means, confidence intervals (CIs), and coefficients of variation (CVs) of the analysed mineral elements and isotopic ratios in the apple fruit flesh of ‘Golden Delicious’ in 2020–2022.

Locality

P (±CI)

CV

K (±CI)

CV

Mg (±CI)

CV

Ca (±CI)

CV

Fe (±CI)

CV

Mn (±CI)

CV

CR-CeB

703 (513–893)

13.19

8485 (6889–10,080)

9.18

338 (286–389)

7.40

261 (214–308)

8.77

7.75 (5.99–9.52)

11.11

4.35 (2.51–6.19)

20.64

CR-EaB

668 (574–762)

6.89

8009 (7222–8796)

4.79

310 (285–334)

3.87

260 (235–284)

4.58

6.98 (6.15–7.80)

5.80

2.89 (2.25–3.53)

10.80

CR-SoM

605 (478–731)

10.21

6808 (5816–7800)

7.11

319 (281–356)

5.74

205 (176–233)

6.78

6.51 (5.37–7.66)

8.60

3.47 (2.33–4.61)

15.99

PL-LoV

467 (369–564)

10.21

7443 (6359–8527)

7.11

279 (247–312)

5.73

244 (210–278)

6.80

6.45 (5.31–7.59)

8.60

2.64 (1.78–3.51)

16.02

PL-LSV

457 (362–553)

10.22

7254 (6198–8311)

7.11

290 (256–324)

5.72

260 (224–296)

6.81

6.28 (5.18–7.39)

8.61

3.78 (2.54–5.02)

16.01

PL-MaV

574 (454–695)

10.23

6824 (5830–7819)

7.11

315 (278–352)

5.71

274 (235–312)

6.79

6.88 (5.67–8.10)

8.60

3.60 (2.42–4.78)

16.00

Locality

Zn (±CI)

CV

B (±CI)

CV

Cu (±CI)

CV

10B/11B (±CI)

CV

87Sr/86Sr (±CI)

CV

CR-CeB

2.57 (1.38–3.76)

22.57

17.4 (10.5–24.4)

19.48

3.00 (1.89–4.11)

18.07

0.193 (0.181–0.206)

3.26

21.10 (6.29–36.00)

34.36

CR-EaB

1.58 (1.20–1.96)

11.77

22.0 (17.4–26.6)

10.18

2.16 (1.74–2.58)

9.44

0.208 (0.201–0.215)

1.70

13.10 (8.31–17.90)

17.94

CR-SoM

1.30 (0.83–1.76)

17.46

21.7 (15.0–28.4)

15.12

1.90 (1.36–2.45)

14.05

0.200 (0.189–0.210)

2.52

23.20 (10.59–35.90)

26.59

PL-LoV

1.53 (0.98–2.08)

17.52

15.9 (11.0–20.8)

15.09

1.17 (0.83–1.50)

14.02

0.186 (0.176–0.195)

2.52

11.70 (5.34–18.10)

26.58

PL-LSV

1.88 (1.21–2.55)

17.45

19.2 (13.3–25.1)

15.05

1.78 (1.27–2.29)

13.99

0.191 (0.182–0.201)

2.52

27.00 (12.31–41.70)

26.56

PL-MaV

1.52 (0.98–2.07)

17.50

20.0 (13.8–26.1)

15.05

1.93 (1.38–2.49)

14.04

0.213 (0.202–0.224)

2.52

18.70 (8.54–28.90)

26.63

Aside from the initial model for PCA and LDA comprising all studied mineral elements and isotopic ratios, two additional models were analysed. The goal was to improve the potential discrimination rate among the samples from different districts. This was done based on the results of the ANOVA test of particular variables for each cultivar. This led us to a model with P, K, Ca, Cu, and ¹⁰B/¹¹B for apple cultivar ‘Gala’ and a model with P, Ca, Cu, and ¹⁰B/¹¹B for ‘Golden Delicious’. Third model was built to comprise only the analysed isotopic ratios of strontium and boron.

All the statistical analyses were performed in “RSstudio, version 2023.06.1” with “R” software, version 4.3.0, with the use of the libraries Factoextra [24], MASS [25], and ggplot2 [26].

3. Results

The contribution from the whole set of mineral elements and isotopic ratios analysed in this study was high for fruit flesh of apple cultivar ‘Gala’ with P, Mg, B, Fe, and ¹⁰B/¹¹B, and low especially with Ca and Zn (Figure 1A). The PCA diagram (Figure 1D) shows that the first two components describe only 48.1% of the total variance of the data containing all the analysed mineral elements and isotope ratios. The first component was mostly correlated with elements P, Mg, and Fe, while the second component shows high correlation with boron and its isotopic ratio ¹⁰B/¹¹B (Figure 2A). Moderate correlation with the first component was also found with K, Mn, and Cu. On the other hand, the mineral elements with the highest contribution to separation of the samples within the first two components do not always contribute to the separation of the samples among the analysed districts. The coefficient of variation across all samples of cultivar ‘Gala’ was moderately high with ⁸⁷Sr/⁸⁶Sr (68.8%), Mn (46.9%), Zn (35.5%), Cu (28.7%), and Ca (25.1%), then moderately low with P (17.3%), Fe (17.1%), Mg (13.2%), and K (12.8%), and low with ¹⁰B/¹¹B (7.6%). The variability of the mineral content among the different districts was uneven with Mn, Zn, and the isotopic ratios of ¹⁰B/¹¹B and ⁸⁷Sr/⁸⁶Sr according to the Cochran–Hartley–Bartlett test (for more details, see the CV in Table 1). However, while Mn (14.4–23.5%), Zn (10.4–17.0%), and ⁸⁷Sr/⁸⁶Sr (19.2–31.4%) were characterized by relatively moderate coefficients of variation within a particular district, ¹⁰B/¹¹B showed just very low values between 2.3–3.8%. The coefficients of variation were higher in Central Bohemia (CeB) and South Moravia (SoM) districts whereas they were lower, especially in Eastern Bohemia (EaB) and Lower Silesian Voivodeship (LSV). According to the mineral profile of the apple fruit, samples from CeB were separated from samples from almost all analysed districts except for SoM. The most distant districts were the LSV and Łódź Voivodeship (LoV). The separation between CeB and LSV was achieved mainly by a trend towards a lower content of P. Aside from the similar trend with P, the samples collected in CeB contained significantly higher Cu than those found in LoV (F(5,24) = 4.084, p = 0.008**; Table 1). Another separation can be tracked between the samples of cultivar ‘Gala’ fruit coming from Masovian Voivodeship (MaV) and LoV in P, Cu, and B, including the analysed boron isotopic ratio ¹⁰B/¹¹B (F(5,24) = 3.013, p = 0.030*). The difference between the mineral content of fruits from MaV and LSV were close to being statistically significant for K (F(5,24) = 2.321, p = 0.075), but the apples still showed overlap between the confidence intervals of the two groups (Figure 1D). Another significant difference was found in Ca content in ‘Gala’ fruit flesh between CeB with 449 mg Ca.kg⁻¹ of DW (CI 341–558 mg Ca.kg⁻¹) and East Bohemia (EaB) with 460 mg Ca.kg⁻¹ of DW (CI 392–528 mg Ca.kg⁻¹), showing higher values in comparison with samples analysed from SoM (284 mg Ca.kg⁻¹ of DW; CI 215–352 mg Ca.kg⁻¹; F(5,24) = 3.423, p = 0.018*). However, because of the relatively small contribution of the Ca content to the first two components, this difference is hardly able to be tracked in Figure 1D. The results of the LDA show that the mineral composition of fruit flesh of apple cultivar ‘Gala’ with the full set of analysed elements gives five linear discriminants (LD): LD1, LD2, LD3, LD4, and LD5. These LDs explained 53.30%, 34.42%, 8.24%, 2.74%, and 1.30% of the given variability, respectively. However, the prediction rate of the linear discriminant model was only 50.0%.

The mineral elements in the reduced PCA (Figure 1B) show the highest contribution with K and P, and the lowest with Ca. Here, the first two PCA components explain 62.3% of the observed variability (Figure 2B). However, the next two components remained also important with a contribution within the range of 10–20%. The results of the reduced PCA (Figure 1E) show only slight differences against the full set of elements (Figure 1D). The only two differences were observed in eliminating the overlap between MaV and SoM and a new overlap between MaV and CeB (Figure 1E). In the first case, the difference between the samples from MaV and SoM could occur with the exclusion of Mg, which was very similar among these samples. In the second case, the occurrence of overlap could be caused by the exclusion of Sr isotopes. The results of linear discriminant analysis with the reduced model also showed only limited prediction rate corresponding to 40.0% of correctly classified samples.

If the isotopic ratios for boron (¹⁰B/¹¹B) and strontium (⁸⁷Sr/⁸⁶Sr) are considered alone (Figure 1C), some differences among the groups of samples with different origins were found (Figure 1F). The MaV expressed a higher ¹⁰B/¹¹B ratio than LoV, but a certain trend can also be followed with CeB and EaB, which were able to be distinguished from LoV and, in case of CeB, also in comparison with MaV and EaB, which had the highest ¹⁰B/¹¹B ratio. The linear discriminant analysis did not bring a model with higher prediction rate (40.0%) than either the original or reduced models.

Among the whole set of mineral elements and isotopic ratios, a higher contribution to the distribution of samples with fruit flesh of apple cultivar ‘Golden Delicious’ was found with P, Mg, Mn, Fe, and Zn (Figure 3A). The lowest contribution was found with calcium and both analysed isotopic ratios. The PCA diagram (Figure 3D) shows that the first two components describe only 48.4% of the total variance of the data containing all the analysed mineral elements and isotope ratios. The coefficient of variation across all samples of cultivar ‘Golden Delicious’ was moderately high with ⁸⁷Sr/⁸⁶Sr (67.1%), Zn (43.9%), Mn (40.9%), Cu (39.1%), B (37.5%), and P (25.7%), then moderately low with Fe (17.1%), K (16.4%), Ca (16.1%), and Mg (13.1%), and low with ¹⁰B/¹¹B (7.2%). The variability of the mineral content among the different districts was uneven with Mn, Cu, and the isotopic ratio of ¹⁰B/¹¹B according to the Cochran–Hartley–Bartlett test (for more details, see the CV in Table 2). The coefficients of variation were overall higher in the CeB than in the EaB district. The clustering of the samples shows just very little differences among the analysed districts. The separation between LoV and both EaB and MaV is the result of a significant difference in the ¹⁰B/¹¹B ratio (F(5,28) = 4.879, p = 0.002**). The samples from the EaB district (668 mg P.kg⁻¹ of fruit flesh DW; CI 574–762 mg P.kg⁻¹) show significantly higher P content than LoV, having 467 mg P.kg⁻¹ (CI 369–564 mg P.kg⁻¹), and LSV, having 457 mg P.kg⁻¹ of DW (CI 362–553 mg P.kg⁻¹; F(5,28) = 3.286, p = 0.019*). Samples from LoV also showed lower content of Cu (1.17 mg Cu.kg⁻¹; CI 0.83–1.50 mg Cu.kg⁻¹) than those coming from CeB, with 3.00 mg Cu.kg⁻¹ (CI 1.89–4.11 mg Cu.kg⁻¹), and EaB, 2.16 mg Cu.kg⁻¹ (CI 1.74–2.58 mg Cu.kg⁻¹; F(5,28) = 4.101, p = 0.006**). A certain trend towards lower Ca content was observed in SoM when compared with samples from the MaV and EaB districts, but the differences were not significant (F(5,28) = 2.369, p = 0.065). The results of the LDA show five LDs explaining 87.43%, 8.92%, 2.58%, 0.01%, and less than 0.01% of the given variability, respectively. However, the prediction rate of the linear discriminant model was very low, having an accuracy of only 15.4%.

The contribution of mineral elements in the reduced PCA (Figure 3B) shows the highest contribution of Ca and the lowest with ¹⁰B/¹¹B ratio. Here, the first two PCA components explain 69.0% of the observed variability (Figure 4B), but again, the third PCA component remained relatively important with a contribution higher than 20%. The results show that the reduced PCA improves the difference among the analysed districts (Figure 3E). The clusters of EaB and MaV became more distant from the districts LoV and LSV. Additionally, the SoM district formed an almost independent PCA cluster. However, the prediction rate of the linear discriminant model still remained very low, having an accuracy of only 26.3%.

The analysis of the isotopic ratios (Figure 3C) for boron and strontium allowed the differentiation of two main clusters containing, on the one hand, samples of CeB, LoV, and LSV, and on the other hand, EaB and MaV. The samples of SoM were distributed across both mentioned clusters. The prediction rate of the linear discriminant model was the highest among the models for ‘Golden Delicious’, but still low at 30.8%.

4. Discussion

The mineral and isotopic analysis of the fruit flesh of apple trees provided interesting insights for the potential classification of their geographical origin. In this context, particular elements showed different potentials for the discrimination of geographical origins. The results of the variation coefficient show a relatively high distribution of single elements and isotopic ratios, particularly of ⁸⁷Sr/⁸⁶Sr, manganese, zinc, copper, calcium, phosphorus, and boron. This gives them relatively high potential for appropriate separation among the analysed districts. On the other hand, their frequent, relatively high variation within a particular district complicates the geographical separation of the samples, showing significant differences with Ca and Cu only. For mineral elements, the obtained variance is comparable with results from apple fruit analyses in previous studies [27,28], where the variability of Ca, Fe, Mn, Zn, but sometimes also K remains relatively high. However, at least some of the analysed elements usually allow discrimination among different geographical areas for both fruits [10,11,28] or products like fruit juices [12,21]. The possible difference could be linked to the portfolio of the analysed mineral elements.

Relatively low variation among analysed samples was found in potassium, magnesium, and ¹⁰B/¹¹B, but only the boron isotopic ratio provided some differences among the analysed districts. Thus, through the comparison of isotopic ratios, the isotopic ratio of boron (¹⁰B/¹¹B) appears to be more useful in the evaluated areas compared to the commonly recommended ⁸⁷Sr/⁸⁶Sr ratio of strontium isotopes [8]. However, as the analysis of Sr isotopic ratio by ICP-MS was recently found to provide insufficient resolution [29], we are aware that the relatively small differences in isotopic signatures of the geochemical soil composition involved, and datasets measured on the used instrument (single-collector ICP-MS) may not be sufficient evidence to connect the fruit-soil-bedrock signatures. Our results show thus only a relatively low partial discrimination of the samples among different districts.

There are multiple factors which may counteract the results of the mineral profiles of apple fruit obtained in this study, which increase their variability for a particular district. In principle, the mineral content of fruit is affected by the balance between its uptake and utilization. Here, the main source of mineral nutrients for plant uptake is the soil. Due to the significant dependence of the content of elements in the fruits and the soil properties on different localities [8], it can be stated that the soil conditions in some growing areas in Eastern Bohemia, but also South Moravia, are probably close to the soil composition in some areas in Poland. Overlaps can be found especially in the occurrence of Arenosols, Cambisols, and Luvisols [30,31]. The reason may be the similarity in the composition of the soil elements due to pedogenesis as a result of complex geological and tectonic changes in Central Europe [32,33]. On the other hand, the Bohemian Massif was described geologically as very heterogeneous [34], which could correspond with increased heterogeneity of the sampled mineral content and isotopic ratios inside the regions, especially Central Bohemia.

It was also surprising that, despite the comparable conditions of phosphorus content in the upper layers of the soil at the Polish and Czech sites [13], the phosphorus content in apples was consistently a significant element differentiating these clusters. On the contrary, despite the generally low content of potassium in Polish soils in comparison with the Czech soils, this element did not represent a significant component in differentiating the resulting clusters. As most of the elements analysed in this work serve as mineral nutrients for the fruit trees, their natural content in the fruit flesh may be influenced by management of fruit orchards, especially soil management, fertilization, and water regime of the trees [14,16,34,35], used rootstock/cultivar combination [18,36], and plant protection, where copper enters as a component of the active compound [37]. Furthermore, the mineral composition of fruits can be changed by the trees’ performance, i.e., their growth and fruit load [4]. However, only a detailed survey could shed more light on the portion of the variability in mineral element content caused by plant development and anthropogenic intervention. From a practical point of view, our study and its results at this stage allow imitating the possibilities of food inspection authorities in particular countries for the authentication of the country of origin of fruits on the market. In this case, they would not have information on agro-technical measures at each site.

As the best markers for the separation among particular districts in the regions of Bohemia, Moravia, and South and Middle Poland seem to be the content of P, Ca, Cu and the isotopic ratio of ¹⁰B/¹¹B. However, the resolution of the geographical discrimination needs to be improved before it can be used for the purposes of laboratory authentication of the fruit origin. Possible ways for this improvement are to characterize the districts using an enhanced set of primers, to use more precise analyses of Sr isotopic ratio via TIMS or MC-ICP-MS techniques, or to select combined multiple chemical analyses. Furthermore, the geographical resolution of mineral analyses can change with the proximity of the analysed regions [12], giving the best results among regions with different geological origins [21]. To tackle the heterogeneity in more geomorphologically variable regions, it could be advantageous to perform the analyses further into micro-regions.

5. Conclusions

In conclusion, it can be stated that the first achieved results of the analysis of the mineral composition and isotopic ratios of Polish and Czech apples show certain trends, especially for the elements P, Ca, Cu, and isotopic ratios of boron. However, these trends alone do not allow a sufficient distinction of the geographical origin of apples, and further research in this area is needed. Due to relatively high intra-regional variability, most biogenic macro- and microelements, especially K, Mn, Fe, and Zn, are less reliable for geographical authentication of the fruit origin because of their relation to the agronomic management and tree performance, likely altering their natural occurrence in the evaluated fruit orchards. Further research needs to be done to perform an extended mineral analysis with an extended set of elements or a combined multidimensional analysis with other methods, such as the evaluation of selected phenolic metabolites in the fruit flesh. The heterogeneity of the analysed districts could be also tackled by more detailed analyses covering the geographical heterogeneity through the analysis of the element content along distinct micro-regions rather than geopolitical regions only.