Analysis of δ¹³C and δ¹⁵N values in Croatian honey by EA–IRMS and possibility of their application in botanical origin verification

Abstract

The aim of this work was to give characteristic stable carbon and nitrogen isotope ratio (δ¹³C_honey, δ¹³C_protein and δ¹⁵N) ranges and examine their relation with botanical origin of honey. Despite that δ¹³C parameter has primary purpose to detect honey adulteration, stable isotopes generally have become important parameter for detection its botanical and geographical origin. The data about stable isotopes are scarce in comparison to other well-known parameters in honey, and in Croatia there is no data about stable isotopes in unifloral honey. This research includes six characteristic honey types (black locust, chestnut, lime, rape, winter savory, and sage honey) from Croatia. Large number of differences between honey types were found in the analyzed IRMS parameters. PCA analysis has successfully separated winter savory from all other honey types, except sage honey, whose samples differed from black locust samples.

Supplementary Information

The online version contains supplementary material available at 10.1007/s13197-023-05888-9.

Introduction

Honey is a sweet natural food with chemical composition predominated by carbohydrates, mainly glucose and fructose (65–75%) and water (15–20%) (Vincevica-Gaile et al. 2011). Beside carbohydrates and water, there are components that are present in small quantities (free amino acids, proteins, enzymes, vitamins, minerals, phenolic compounds, flavors and aromas etc.), which are responsible for the sensory and nutritional properties of honey (Silva et al. 2009). Physicochemical composition of honey is affected with its botanical and geographical origin (Resende Ribeiro et al. 2015), but also with climatic and environmental conditions, so as beekeeping practice (Juan-Borrás et al. 2014; Silva et al. 2009). Numerous studies over the last thirty years have focused on the characterization of honey in order to find markers for its botanical and geographical origin determination. Both characteristics are related to consumer demands and the price of honey. Honey of a particular plant species that comes wholly or mainly from one source and possesses sensory, physicochemical and microscopic properties of that source (unifloral honey) is generally more expensive than multifloral, and some types are several times expensive. The situation is similar with geographical origin. In Western Europe, honey originating from the Far East and Latin America has a lower price than locally produced honey, and there are differences between countries and regions within the European Union (Ruoff and Bogdanov 2004). All of the above is a good basis for honey adulteration. Honey analytical methods are trying to meet the requirements for botanical and geographical origin identification and considerable efforts are being made to find new methods to solve this problem. Well-known physicochemical and melissopalynology analyses are becoming insufficient due to increasingly different and innovative ways of honey adulteration, whether it is an incorrect declaration of botanical or geographical origin, or more and more represented addition of sugar syrups. Some more recent techniques used to confirm the authenticity of honey or other bee products are: inductively coupled plasma–mass spectrometry (ICP–MS), gas chromatography (GC), gas chromatography–mass spectrometry (GC–MS), liquid chromatography–mass spectrometry (LC–MS), infrared spectroscopy (IR), nuclear magnetic resonance (NMR), etc. One of the newer analytical approaches is the use of isotope ratio mass spectrometry (IRMS) together with multivariate statistical methods whose result allow proving the authenticity of honey (Dinca et al. 2015a, b; Milojković Opsenica et al. 2015). The primary purpose of stable carbon isotopes ratio determination in honey is to detect adulteration via the δ¹³C (¹³C/¹²C) isotope ratio. The reason for the increasing adulteration of honey lies in the fact that honey is a valuable food whose production requires a lot of time, effort and cost. Adulteration is most often carried out by adding sugar syrups (high fructose corn or cane sugar). Method called stable carbon isotope ratio analysis (SCIRA) was first published by White and Doner (1978) where stable carbon isotopes ratio in honey was expressed as a δ¹³C value and represents the ratio between stable carbon isotopes in honey and the material used as a reference (V-PDB) (Croft 1987). Later, White (1992) revised the imperfection of previous method by adding honey proteins as internal standard (Internal Stable Carbon Ratio Analysis method-ISCIRA). Beside its main purpose is detection adulteration, it is important to know characteristic ranges of stable isotopes in honey given that this is one parameter whose knowledge of characteristic ranges could contribute to honey authenticity. Many countries have already attributed the characteristic stable isotope ranges to their most common honey types (Bontempo et al. 2017; Daniele et al. 2012; Dinca et al. 2015a, b; Kropf et al. 2010; Pang et al. 2006) in the recent years and some of the authors tried to find correlation between stable isotope ratio and botanical origin of honey. Kropf et al. (2010) did not get a clear difference between honey types according the determined stable isotopes with explanation that honey in Slovenia is always a mixture of nectar and/or honeydew honey of different C₃ plants. Bontempo et al. (2017) concluded that botanical origin is clearly influencing factor on isotopic characteristics of honey. Wu et al. (2015) find stable carbon isotopes not suitable for botanical origin determination of Chinese honey, while oxygen and hydrogen isotopes were indicated as good biological marker for discrimination of botanical origin. Schellenberg et al. (2010) proved and confirmed that the ratio of stable carbon isotopes is also influenced by weather factors and that they can also be used as indicators of the geographical origin of honey. In addition to the determination of stable carbon isotopes, nitrogen isotopes are also used to determine geographical origin. The nitrogen isotope ratio δ¹⁵N (¹⁵N/¹⁴N) in isolated honey proteins reflects the soil condition of the area where bees collected nectar, but also of botanical origin, given that some plants (black locust, clover) fix nitrogen from the air. The possibility for distinction of botanical and geographical origin is higher with combination of multivariate analysis and more parameters like IRMS and mineral composition (Bontempo et al. 2017; Wu et al. 2015). In recent years there are researches conducted on Croatian honey with aim to prove its botanical or geographical origin which were based on honey characterization according to physicochemical composition, sensory analysis, melissopalynology, antioxidant capacity, mineral composition, flavonoid profile, etc. (Bilić Rajs et al. 2017; Flanjak et al. 2016; Gašić et al. 2015; Kenjerić et al. 2006; Primorac et al. 2011, 2013). Only available information about IRMS parameters is given in research of Vasić et al. (2020) including δ¹³C in honeydew honey. Still there is no information about stable isotope ratio of most common Croatian honey types which lead to the aim of this work. Based on the previous researches on stable isotope ratio in honey, hypothesis of this work was that stable carbon and nitrogen isotope ratio can be used for honey botanical origin determination. In this study characteristic ranges of carbon and nitrogen stable isotopes were determined in six groups of unifloral honeys (black locust, chestnut, lime, rape, winter savory, sage) produced in Croatia. Furthermore, relation between stable isotope ranges and botanical origin were tested, so as influence of production year and compliance of carbon ratios with the official AOAC limit (AOAC 998.12 2000).

Materials and methods

Honey samples

The research was conducted on 144 honey samples belonging to six different botanical sources: black locust (Robinia pseudoacacia L.) N = 46, chestnut (Castanea sativa Mill.) N = 26, lime (Tilia spp.) N = 26, rape (Brassica spp.) N = 21, winter savory (Satureja montana L.) N = 11, sage (Salvia officinalis L.) N = 13 collected on the territory of the Republic of Croatia from the beekeepers with a denominated origin. Samples of black locust, chestnut, lime and rape honey were produced and collected in the both harvesting seasons 2014 and 2015, while winter savory and sage honey samples were produced and collected in one harvesting season (2014 and 2015, respectively). Honey samples were placed in the glass bottles, labeled and placed in the dark room until analyses. For botanical origin verification physicochemical (Table S1) (Codex Standards 12-1981 2001; International Honey Commission 2009 and The Council of the European Union 2002), melissopalynological (DIN 10760 2002) (Table S1) and sensory (Piana et al. 2004) analyzes were conducted following the requirements of Regulation on the quality of honey (Ministry of Agriculture 2015), Regulation on the Quality of Unifloral Honey (Ministry of Agriculture, Fisheries and Rural Development 2009) and Council Directive 2001/110/EC relating to honey (The Council of the European Union 2002).

Sample preparation for stable carbon and nitrogen isotope ratio analysis

Honey samples for stable carbon and nitrogen isotope ratio were prepared for determination according to AOAC method (AOAC 998.12 2000) in duplicate. Before the analysis on EA-IRMS, proteins were isolated from honey samples using distilled water, 0.335 M sulphuric acid (Carlo Erba, France) and 10% sodium tungstate solution (Kemika d.d., Croatia). Prepared solutions were thermostatically controlled at 80 °C until visible floccules appeared and afterwards centrifuged at 1500 g (Sigma 2–16, Germany). Sediment was dried in oven (NUVE FN 400P, Turkey) and freezed in dark glass container until analyses. 0.2 mg of honey/protein was weighted with microbalance (MX-5 ultramicrobalance, Mettler-Toledo, Germany) into a small tin container (IVA Analysentechnik, Germany). The container was closed using tweezers.

Stable carbon and nitrogen ratio analysis

Prepared capsules were placed in a solids autosampler (MAS 200R, Thermo Scientific, UK) connected with EA-IRMS (FlashEA 2000 HT with NoBlank device–Thermo-Electron Delta V Advantage Isotope Mass Spectrometer, Thermo Scientific, UK) device via ConFlow IV Interface (He pressure: 0.8 bar, CO₂ pressure: 2 bar). Elemental analyzer was operating in NC modus and controlled with Isodat 3.0 software. EA conditions were: reactor temperature 1020 °C, GC temperature 60 °C, carrier flow 100 mL/min, O₂ flow 250 mL/min, O₂ injection time 1 s, autosampler delay 16 s. Reference standard for carbon was PeeDee Belemnite (PDB) (AOAC 998.12 2000). CO₂ and N₂ reference gas pulse was introduced two times (20 s each) at the beginning of each run. The ConFlow Interface diluted the CO₂ sample peak with He in a split of about 1:10 and N₂ sample peak was not diluted. The Delta V Advantage (IRMS) mass spectrometer with triple ion collector, was used to simultaneously measure the signals at m/z 44, 45, and 46 of the molecular ions of the CO₂, and the signal m/z 28, 29, 30 of the molecular ions of the N₂, formed by the sample combustion.

The isotopic value of the honey/protein sample (δ¹³C) is compared with the international standard Pee Dee Belemnite (PDB), through Eq. 1, where R_sample is the ¹³C/¹²C isotopic ratio of the sample and R_ref is value of the international standard PDB (AOAC 998.12 2000).

$$\delta =\left[\left(\frac{R_{\mathit{sample}}-R_{\mathit{ref}}}{R_{\mathit{ref}}}\right)\right]\ast 1000\left[‱\right]$$ 1

The CO₂ reference gas used to calculate the δ¹³C values in each analytical run was calibrated with a laboratory working standard all purchased from Sigma Aldrich (glucose monohydrate − 26.990 ± 0.2‰; acetanilide − 27.26 ± 0.2‰ and aspartic acid − 21.67 ± 0.2‰) which were calibrated against Community Bureau of Reference (BCR) the stable carbon reference standard (δ¹³C sugar in ‰ VPDB with certified value − 10.76 ± 0.04‰). The N₂ reference standard used to calculate the δ¹⁵N values in each analytical run was calibrated with a laboratory working standard (acetanilide 0.99 ± 0.03 ‰ and aspartic acid − 7.56 ± 0.04‰) which were calibrated against stable nitrogen reference standard IAEA-N1 (+ 0.4 ± 0.2‰) and IAEA-N2 (+ 20.3 ± 0.2‰) (International Atomic Energy Agency, Vienna, Austria). The laboratory working standard was measured in each sequence of sample measurements at least 4 times.

Statistical analysis

Results were expressed graphically (mean, median, extreme, minimum, maximum) for all analysed honey types. Kruskall-Wallis test (KW test) was used for honey type discrimination according to botanical origin. In case of rejection of median equality hypothesis for certain honey characteristic (δ¹³C_honey, δ¹³C_protein, δ¹⁵N), Dunn test post-hoc analyses was conducted with Holm correction of p value. Mann–Whitney–Wilcoxon (MWW) was used to test the existence of differences in two production year of collected honey samples. All conclusions were enacted at significance level p ≤ 0.05. Principal Component Analysis (PCA) was used for potential classification of honey according to botanical origin. Statistical analysis were conducted in Statistica 13 (Stat. Soft. Inc., USA), Microsoft Office Excel and R-3.5.2 software.

Results and discussion

Characteristic stable isotope ratio ranges and correlation with botanical origin

Stable isotopes have been used for a long time for detection of adulteration with sugar syrups (δ¹³C), but researches in the recent years show their potential in determination of botanical and geographical origin of the honey. Due to the lack of information about stable isotope ratio in Croatian honey in this research δ¹³C_protein, δ¹³C_honey and δ¹⁵N values in selected honey types were analyzed. The obtained results are depicted in Fig. 1. Black locust honey had δ¹³C_honey range from − 26.46 to − 24.10‰ and it had, in comparison to other analyzed honey types, higher maximal and average value (− 25.17‰) (Fig. 1a). Wide range for δ¹³C_honey parameter could somewise be in relation with pollen share and electrical conductivity of particular sample. Black locust honey had wide range of mentioned parameters (Table S1) and samples with higher share of Robinia pseudoacacia L. pollen and lowest electrical conductivity values showed tendency to have more negative δ¹³C_honey values. In opposite, samples with lower share of pollen and higher electrical conductivity values had less negative δ¹³C_honey values. Black locust honey sample with highest δ¹³C_honey value (− 24.10‰) had electrical conductivity 0.21 mS/cm and pollen share 24%, while sample with lowest δ¹³C_honey value (− 26.46‰) had electric conductivity 0.10 mS/cm and pollen share 44%. Chestnut, lime and sage honey had very similar average values (− 26.14‰, − 26.12‰ and − 26.44‰, respectively), while the lowest average values had the rape and winter savory honey (− 27.41‰ and − 27.93‰, respectively) (Fig. 1a). By testing the statistically significant difference of δ¹³C_honey values, in accordance to honey types, black locust honey differs from other honey types while chestnut honey and lime honey differ, beside the black locust honey, also from rape and winter savory honey (Table 1). Also, a statistically significant difference exists between winter savory and sage honey (Table 1). Some authors (Dinca et al. 2015a, b; Kropf et al. 2010) have shown that the examined parameter is not suitable for distinguishing botanical origin of honey. They start from the fact that most honey plants belong to the C3 group, so the real difference within the δ¹³C in honey and δ¹³C in honey proteins is not to be expected. Although mentioned authors obtained differences in the mean values for these parameters in their study, the ranges for individual species overlapped. Dinca et al. (2015a, b) and Kropf et al. (2010) obtained the highest δ¹³C_honey values in black locust honey (− 23.14‰ and − 24.8‰) and the lowest values in rape honey (− 25.72‰) (Dinca et al. 2015a, b) who had also, together with winter savory honey (− 27.93‰), the lowest value (− 27.41‰) in this research. The research carried out by Bontempo et al. (2017) agrees with the results of this study, they successfully distinguished black locust and chestnut honey based on δ¹³C and δ¹⁵N values. All δ¹³C literature values for black locust, chestnut, lime and rape honey were slightly higher than those obtained in this study (Dinca et al. 2015a, b; Daniele et al. 2012; Kropf et al. 2010; Pang et al. 2006) and the difference was around 1‰. For winter savory and sage honey these are the first available data about stable isotope ratio.

Fig. 1.

Display of aggregate values for δ¹³C_honey, δ¹³C_protein and δ¹⁵N_protein in honey (x average; - median; o extreme; ┴ ┬ min and max that does not include extremes, □ Q1–Q3)

Table 1.

Difference in δ¹³C and δ¹⁵N values according to honey type and production year

Parameter	δ13Choney							δ13Cprotein							δ15Nprotein
Honey type	BL	C	L	R	WS	S		BL	C	L	R	WS	S		BL	C	L	R	WS	S
BL		+	+	+	+	+	a		+	+	+	+	+	a		+	+		+	+	a
C	+			+	+		a	+			+				+			+	+	+	a
L	+			+	+			+						a	+			+	+	+	a
R	+	+	+				a	+	+					a		+	+		+	+
WS	+	+	+			+		+							+	+	+	+
S	+				+			+							+	+	+	+

BL—black locust, C—chestnut, L—lime, R—rape, WS—winter savory, S—sage

+ There is a statistically significant difference regard to the type of honey according to the KW test (p ≤ 0.05)

a There is a statistically significant difference between production year according to the MWW test (p ≤ 0.05)

According to Schellenberg et al. (2010) weather conditions like number of sunshine days, precipitation and relative air humidity affects δ¹³C_protein values so this parameter could also be used for determination of geographical origin of honey. Clearly different from the others was black locust honey which had the highest mean value (− 25.14‰) (Fig. 1b), while the ranges and mean values of other honey types were similar (chestnut − 26.23‰, lime − 26.44‰, rape − 27.02‰, winter savory − 26.65‰ and sage − 26.66‰) (Fig. 1b). Testing the statistically significant difference of δ¹³C_protein values showed that black locust honey differed from the other honey types, and chestnut honey from rape honey (Table 1). The data for French rape honey (− 26.3‰) (Daniele et al. 2012) is higher than the results obtained in this study. The mean values obtained by Bontempo et al. (2017), Chen et al. (2019), Daniele et al. (2012), Kropf et al. (2010), Pang et al. (2006) and Simsek et al. (2012) for black locust, lime and chestnut honey were also higher. Schellenberg et al. (2010) in their study obtained a median range of δ¹³C _protein values of black locust honey from − 26.1‰ (Germany) to − 24.2‰ (France) which is a range of about 2‰ for different regions in Europe, but on the other hand, when they compared the value differences of δ¹³C_protein in five different types of honey within one area (France), obtained difference was 1.8‰. It can be seen from the above that differences between (some) species exist, and they can also affect the differences of δ¹³C of honey proteins from different areas. Vasić et al. (2020) concluded that the values of δ¹³C_protein isolated from the honeydew honey could be a good botanical origin indicator.

Nitrogen is essential element for plant metabolism, and given how each plant has adapted to this element (e.g., heather Caluna spp. grows on nitrogen-poor soils), δ¹⁵N in honey proteins is also expected to reflect the botanical origins of honey (Schellenberg et al. 2010). δ¹⁵N also reflects the soil composition of the area where the bees collected nectar, given that amino acids and proteins in plants are formed from soluble substances in the upper layer of the soil where plants grow (Schellenberg et al. 2010). The mean values of δ¹⁵N in winter savory and sage honey were − 0.10‰ and 0.95‰ (Fig. 1c), respectively, which is much lower than other analyzed honey types. Rape honey, whose all samples originate from continental region of Croatia, stands out with the highest δ¹⁵N mean value (4.02‰) (Fig. 1c). Similar average value was reported in the work of Labsvards et al. (2021) where δ¹⁵N value for rape honey was 4.9‰. Chestnut and lime honey had similar mean values (2.64‰ and 2.65‰, respectively), while the values for black locust honey were higher (3.53‰) (Fig. 1c). Higher average values for lime honey were reported in the work of Labsvards et al. (2021) and Czechowska and Wierzchnicki (2013) (5.8‰ and 4.1‰, respectively) while Dinca et al. (2015a, b) reported similar values for lime (2.57‰) so as chestnut honey (2.05‰) from Romania. Testing the statistically significant difference of δ¹⁵N values with respect to honey type showed that black locust and rape honey do not differ, while all other types of honey differ from them (Table 1). In the literature so far, there is no data on δ¹⁵N in rape, sage and winter savory honey. Based on the obtained results, the set hypothesis was partially accomplished.

The classification of honey according to botanical origin was carried out regard to potentially predictive variables determined by preliminary analysis. PCA analysis included all honey samples collected in both years. The main components of PC1 and PC2 were a function of IRMS variables (δ¹³C_honey, δ¹³C_protein and δ¹⁵N_protein) (Fig. 2), the main components explained 89.81% of the total variability among the data. Classification was not successful due to large data overlaps. Namely, winter savory honey was separated from all honey types except sage honey, and black locust honey was separated from sage honey, beside mentioned winter savory honey (Fig. 2). Wang et al. (2022) concluded in his work that due to complex composition of honey which is influenced by numerous factors better discrimination of honey botanical origin could be achieved with combination of several parameters together with advanced statistics which is the plan for further research in this field.

Fig. 2.

Classification of honey samples according to botanical origin using PC1 and PC2 as functions of IRMS variables

Authenticity according to limit indicated by AOAC

Determination of δ¹³C in honey is primary used for detection of adulteration by addition of sugar syrups originated from C4 plants (maize and cane) where δ¹³C_protein is used as an internal standard for comparison with δ¹³C_honey according to AOAC method (998.12 2000). Honey can also be adulterated with addition of sugar syrups originated from C3 plants (e.g. sugar beet) but this method (EA-IRMS) cannot detect this form of adulteration (Zhou et al. 2018). Elflein and Raezke (2008) proposed EA/LC-IRMS method for detection more sophisticated adulterations with syrups produced from both plant sources (C3 and C4) but there were false positive results obtained by using this method so Xu et al. (2020) concluded in their research that the same criterion should not be applied to all honey samples from all over the world because δ¹³C_honey values can be affected by many factors.

Since two parameters that allow detection of adulteration with C4 sugar syrups are determined in this research the number of samples that are suspicious was also calculated. Seven samples had the value of added sugar syrup near marginal 7% (AOAC 998.12 2000) in range of 6.61–6.97%. Bontempo et al. (2017) does not interpret values near 7% as adulterated, they interpret them as authentic due to calculation of method uncertainty. Elflein and Raezke (2008) in their research claim that use of δ¹³C_protein in honey as internal standard could be weak side of this method. Reason could be high measurement uncertainty of samples with low protein content (e.g. black locust or lavender honey) or risk of altered δ¹³C_protein value due to high content of yeast or residues of bee food supplement. False positive tendency of AOAC method was also shown in the work of Akyıldız et al. (2022) where majority of the authentic pine honey samples seemed to be adulterated with C4 sugar syrups. The reason lies on presence of honeydew elements who can form a notable amount of precipitate who consequently could change δ¹³C_protein values inaccuretly. Four of all analyzed samples had value of added sugar syrup higher than 7%. Those were two samples of black locust honey (G5-12%, G6-11%) and two samples of chestnut honey (K3-10% and K5-9%). All four samples did not point to adulteration because they had physicochemical characteristics in accordance to Croatian regulation (Ministry of Agriculture 2015) and they were in ranges given by Persano Oddo and Piro (2004). Also, beekeepers from whom the samples were collected are known for their good beekeeping practice.

Influence of year of production on stable isotope range

In order to obtain representative data and to examine the influence of production year and weather characteristics on the stable isotope ratio, honey samples were collected over two production years (except sage and winter savory honey). A significant difference in δ¹³C of honey regarding year of production was recorded in all honey types except lime honey (Table 1). The values were lower in 2015 (Fig. 3a) with the exception of rape honey. δ¹³C in honey protein, with the exception of chestnut honey samples where no difference between the two years has been recorded, values were lower in 2015 (Fig. 3b; Table 1). From the data for average monthly temperatures and total rainfall for the flowering period of melliferous honey plants for the regions where the samples were collected (Table 2) it can be seen that, with the exception of black locust, significantly higher precipitation was in 2014, while the differences between temperatures were negligible. As δ¹³C values of honey protein should increase with increasing sunny days, temperature, and decreasing precipitation (Schellenberg et al. 2010), lower carbon isotope values are expected in 2014. The averages of precipitation (142.7 mm and 146.5 mm) and temperature (15.1 °C and 16.9 °C) for the whole of May, in which black locust blooms, did not differ much (Table 2), and the explanation for the higher values obtained in 2014 possibly lies in the fact that the weather conditions in the short flowering periods in a particular region, and even the micro location, differ from the average for the whole month shown in this study. The flowering period in temperate climates is relatively short, while in the Mediterranean climate it lasts much longer. Changes in weather characteristics that occur during these short flowering periods of a particular plant species can significantly affect plant metabolism and thus δ¹³C values (Schellenberg et al. 2010). Influence of the production year and coast effect was also examined in the work of Bontempo et al. (2009) who analysed stable isotopes in extra-virgin olive oil produced trough three years on two locations where precipitation and humidity seemed to be main reason for differences in isotopic ratio in different production years. The results in this study are also affected by the number of samples from each area, such as the case of linden honey samples from 2014, where four samples were from the coastal region with slightly higher δ¹³C values, which possibly affects a higher average data for the whole 2014, compared to 2015 when no sample was from the coastal region. More accurate results would be obtained by multi-year monitoring of δ¹³C values at specific locations and more accurate monitoring of weather characteristics exactly in the flowering days of a particular honey plant, which in this case could not be controlled.

Fig. 3.

Display of aggregate values for δ¹³C_honey, δ¹³C_protein and δ¹⁵N_protein in honey according to production year (x average; - median; o extreme; ┴ ┬ min and max that does not include extremes, □ Q1–Q3)

Table 2.

Average monthly temperatures (° C, dry bulb) and total precipitation for the flowering period of honey bearing plants

Plant species	2014		2015
	Temperature (°C)	Precipitation (mm)	Temperature (°C)	Precipitation (mm)
Black locust	15.1	142.7	16.9	146.5
Chestnut	19.9	113.1	20.5	65.8
Lime	20.1	77.8	20.3	41.9
Rape	13.3	84.2	12.5	18.8

Conclusion

In this study characteristic ranges for stable carbon and nitrogen isotope ratios in Croatian black locust, chestnut, lime, rape, winter savory, and sage honey were determined for the first time. Obtained EA-IRMS data for sage and winter savory honey are the first data generally available for these honey types. It is clear that there are differences between some honey types according IRMS parameters. Black locust honey stood out with highest average δ¹³C_honey and δ¹³C_protein values, while winter savory honey had lowest δ¹³C_honey and sage δ¹³C_protein average value. δ¹⁵N average value was the highest in rape honey, while winter savory and sage honey stood out with lowest one. Despite founded differences, PCA analysis showed that only IRMS parameters were not sufficient for discrimination of honey according to botanical origin due to data overlapping. In the future research, focus should be on more reliable determination of honey botanical origin with inclusion of more parameters. Also, it would be interesting to investigate whether is possible to differ same honey type collected in different regions of Croatia.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

AOAC

Association of official analytical collaboration (AOAC) international

BCR

Community Bureau of reference

EA–IRMS

Elemental analyser–isotope ratio mass spectrometer

EA/LC–IRMS

Elemental analyser/liquid chromatography–isotope ratio mass spectrometer

IAEA-N

International atomic energy agency

ICP–MS

Inductively coupled plasma–mass spectrometry

IRMS

Isotope ratio mass spectrometry

ISCIRA

Internal stable carbon isotope ratio analysis

KW test

Kruskal–Wallis test

LC–MS

Liquid chromatography–mass spectrometry

MWW test

Mann–Withney–Wilcoxon test

PCA

Principal component analysis

PDB

Peedee belemnite

SCIRA

Stable carbon isotope ratio analysis

Author contributions

BBR: Conceptualization; Formal analysis; Writing—original draft; IF: Writing—editing; Visualization; Conceptualization; Methodology; Resource; DMO: Writing—editing; Visualization; ŽT: Writing—editing; FČK: Writing—editing; LJP: Conceptualization; Writing—editing; Visualization; Methodology; Resource.

Funding

Not applicable.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of interest

The authors declare no conflict of interest.