Introduction

The recent COVID-19 pandemic has spurred a heightened interest in the search for food products that can protect against the virus and/or alleviate its symptoms. Although the exact mechanism of anti-viral effect of honey remains unclear, the chemical components of honey, as with many other natural products, are known for their antioxidant, anti-inflammatory and interaction with the structural and non-structural proteins in the virus (Abedi et al. 2021). Hence, it was suggested that honey may prevent COVID-19 virus from entering the host cell, hinder its replication and modulate the inflammatory cascade (Abedi et al. 2021). The pandemic has exhibited rapid growth in honey market demand, with 9.0% in 2020 as compared to 2019. With this, there is anticipated growth in the market from USD 7.84 billion to USD 11.16 billion between 2020 to 2027 (Fortune Business Insights 2021). Honey is a sweet liquid made from nectar that is harvested by bees and used as a natural sweetener in the food industry. It is a complex food, mainly composed of sugary secretion from honeycombs and enriched with other substances such as amino acids, enzymes, minerals, vitamins and organic and phenolic acids. However, due to its high economic value, it is frequently adulterated for profit-making.

Honey is produced and marketed in two main types globally: traditional honey, which is derived from the Apis bee genus, and stingless bee honey (SBH) from the Melipona bee genus. The chemical composition of honey exhibits variability influence by factors such as bee species, the botanical origin of the nectar, climatic conditions and environmental factors. Additionally, the honey’s chemical composition is influenced by factors such as handling, processing method, and storage conditions (Bertoncelj et al. 2007).

Melissopalynology is the study of pollen in honey and is widely used as the least expensive and quickest way to determine the floral or botanical contents and geographical origin of honey (Selvaraju et al. 2019). Through microscopic observation, the morphology of the pollens can be used to identify the botanical origin, based on published pollen slides and photographic atlas as suggested in the harmonized methods of melissopalynology (Von Der Ohe et al. 2004). A honey that contains more than 45% pollen of a specific taxon can be categorized as unifloral honey. Additionally, the geographical origin of the honey could be determined by mapping the entire pollen spectrum in the honey and aligning it with the flora of a specific region (Von Der Ohe et al. 2004). Based on the botanical and geographical information, any potential adulteration or contamination in honey can be detected, and hence, the authenticity and quality of honey can be verified.

A quick search of publications in Science Direct on ‘melissopalynological analysis of honey’ yielded 268 articles in Science Direct and 79 articles in PubMed. These include research articles, review articles, book chapters and short communications. The number of articles showed a growing trend where there were 13 articles on this topic in Science Direct in 2013 and the numbers almost tripled to 37 articles a decade later in 2023. When the search was modified to ‘melissopalynological analysis of stingless bee honey’, only 23 articles had been published so far where two articles published in 2022 and six articles published in 2023. While the information shows growing interest in melissopalynological analysis in general, studies that specifically address SBH are still very limited.

There had been significant research conducted on melissopalynological analysis of honey samples from Apis species, such as Apis cerena (Bhargava 2009), Apis dorsata (Raghunandan and Sekarappa 2014) and Apis mellifera (El Sohaimy et al. 2015), but only limited studies have been conducted on SBH. Most of the melissopalynological analysis of SBH is mainly from Brazil, China and Thailand (Layek et al. 2020; Raja et al. 2014; Yang et al. 2018), but reports on SBH melissopalynological analysis from Malaysia (Majid et al. 2020; Selvaraju et al. 2019), especially in Sarawak, are limited. In Malaysia, stingless bee species are diverse with around 50 species but only two species, Heterotrigona itama and Geniotrigona thoracica, have the commercial potential (Ismail and Ismail 2018), as they are easily discovered in Malaysian forests and domesticated by beekeepers (Shamsudin et al. 2019).

Meanwhile, Sarawak state government emphasizes the need to prioritize environmental sustainability, aligning with Sustainable Development Goals (SDGs) to achieve balance economic growth and provide quality food (Chiam 2022; Ling 2023). Having more information of the composition of local SBH is economically beneficial as it will be able to influence quality control, market value, pricing, research and development, regulatory compliance, and consumer education. Parallel to this, harmonized quality standards, especially for stingless bee honey, are crucial. Given the scarcity of data on its composition, our research aims to serve as a database for international regulations and promote meliponiculture for its health benefits and positive social impacts.

Fourier-transform infrared spectroscopy (FTIR) is a commonly used chemometric approach that is simple, rapid, versatile and cheap for evaluating food constituents (Anjos et al. 2015). Attenuated total reflectance (ATR) is a common sampling technology used with FTIR spectroscopy to examine samples in solid or liquid state without further preparation (Ismail et al. 2021).

The current study aims to identify the principal food sources (botanical origin) of stingless bee in 20 different honey samples mainly from Sarawak, Malaysia via morphological investigation and to categorize the SBH and honey based on their ATR-FTIR spectral data. The melissopalynological information can provide insight into the nutritional qualities of stingless bee honey, based on the botanical origin. This allows producers and entrepreneurs to market honey with unique botanical origins or develop honey-based products with specific attributes. Moreover, categorizing SBH based on ATR-FTIR spectral data can aid in differentiating products in the market, potentially leading to premium pricing for SBH with specific characteristics.

Materials and Methods

Honey Samples

A total of twenty different honey samples, including raw and processed SBH, Apis Acacia honey, and Apis Manuka honey were examined in this study. The raw SBH samples (SBHR1-SBHR12) were purchased based on their availability from SBH suppliers from Kuching, Samarahan and Sibu in Sarawak, Malaysia, whereas the processed SBH samples (SBHP1-SBHP4) were purchased from Sibu, Kuching in Sarawak and from Perak, Selangor, in Malaysia. Both commercial Apis Acacia honey samples (ABH1 & ABH2) are from Sibu, Sarawak. The Apis Manuka honey samples from Australia (Manuka1) and New Zealand (Manuka2) were purchased from local pharmacies.

According to the honey suppliers, all SBH samples were collected from sealed honeypots of Heterotrigona itama stingless bees. The raw SBH samples were minimally processed by filtering them through a cheesecloth or strainer of about 1-mm pore size, and stored in the fridge (~ 5 °C) until analysis. The commercial samples (SBHP, ABH and Manuka) were all subjected to a dehumidifying process to reduce their moisture content to the accepted levels as required by the Codex Alimentarius Honey Standard CX12-1981 (Codex Alimentarius 2019) for Apis honey, and Malaysian Standard for Kelulut (Stingless Bee) Honey MS2683:2017 (Malaysian Standard 2017).

Chemicals

Glacial acetic acid, glycerine and potassium hydroxide were purchased from Sigma-Aldrich (Dorset, UK); gelatin powder from Merck (Darmstadt, Germany); and phenol (C6H5OH) from Amresco (USA).

Melissopalynological Analysis

The melissopalynological analysis of the honey samples was performed using the acetolysis method (Ibrahim et al. 2012; Rosdi et al. 2016). The acetolysis method is commonly used for melissopalynological analysis due to its effectiveness in recovering honey pollen. This method excels in removing debris, lipids and tissues from honey samples and pollen grains, ultimately enhancing the accuracy of pollen identification by making diagnostic characteristics of pollen grains more visible (Jones 2014).

Briefly, 5 mL of honey samples were diluted with 10 mL of warm distilled water, followed by centrifugation at 1200 g for 10 min using an Eppendorf 5810 R refrigerated centrifuge (Germany). After the centrifugation, the supernatant was decanted off, and 3 mL of glacial acetic acid was added to the residue, which was left to stand for 5 min before centrifugation and decantation. The mixture was then centrifuged again at 1200 g for 10 min and the supernatant was decanted. After adding 1 mL of 10% potassium hydroxide to the residue, the mixture was subjected to heating at 70 °C for 5 min. The pollen present in the honey changed colour from light brown to golden brown after the heating process. To eliminate the potassium hydroxide, the mixture underwent another round of centrifugation at 1200 g for 10 min. Subsequently, the resulting pollen residue was uniformly distributed on a microscopic slide coated with glycerine jelly. Finally, the pollen was examined under a Nikon ECLIPSE E100 microscope (USA) with the magnifications of × 400, × 600 and × 1000, and the images of the pollens were captured using a separate camera. Following the captured images, a comparative analysis was performed to associate the honey sample pollens with several literature studies (Rosdi et al. 2016; Yang et al. 2018). This enables the identification of the honey sample’s botanical source for further classification.

Attenuated Total Reflection Fourier-Transform Infrared (ATR-FTIR) Analysis

The ATR-FTIR analysis was performed as described previously by Ismail et al. (2021). The MIRacle ATR accessory in the FTIR spectrophotometer (PerkinElmer, Germany) was used to obtain the infrared spectra in all the honey samples. The honey samples were subjected to spectral measurements within the mid-infrared wavenumber range of 600–4000 cm−1, with 40 interferograms at a resolution of 4 cm−1. Identical instrumental conditions were used to record the background spectrum of the ATR crystal prior to each honey sample analysis. Then, around 1 mL of pure honey sample was loaded in the ATR crystal centre, and the determination of every honey sample was carried out in triplicate. The averages of the FTIR-ATR spectra for all the honey samples were calculated and utilized for the identification of functional groups present in the honey samples.

Then, the spectra were analyzed using PerkinElmer Spectrum™ 3 MIR-FT-IR Spectrometer (USA), and the identified functional groups were compared to the existing literature (Kasprzyk et al. 2018). Absorbance values recorded in selected wavenumbers were then subjected to Principal Component Analysis (PCA), which enables the classifications of the honey samples based on the identified functional groups.

Principal Component Analysis (PCA)

The PCA analysis was conducted as per method of Aitchison (1983) and Sakač et al. (2019), to visualize and explain the discrimination of all the honey samples based on FTIR data. The PCA was computed utilizing well-known Python libraries (scikit-learn, known as sklearn) for the Python programming language.

To compute the PCA, a matrix (Z) with the element (yij) was represented by m number of the honey samples (i) and n number of wavenumber (j), which was denoted in Eq. 1.

$$Z={\left[\begin{array}{cccc}{y}_{\mathrm{0,0}}& {y}_{\mathrm{0,1}}& \cdots & {y}_{0,j}\\ {y}_{\mathrm{1,0}}& {y}_{\mathrm{1,1}}& \cdots & {y}_{1,j}\\ \vdots & \vdots & \ddots & \vdots \\ {y}_{i,0}& {y}_{i,1}& \cdots & {y}_{i,j}\end{array}\right]}_{m\times n}$$
(1)

The dataset of the wavenumber was standardized to achieve a uniform range (0 to 1) across all the wavenumbers. The standardization of the Z matrix was denoted in Eq. 2, where μj and σj denote the average and standard deviation of specific wavenumber (j).

$${\widehat{y}}_{ij}=\frac{{y}_{ij}-{\mu }_{j}}{{\sigma }_{j}}$$
(2)

Then, the covariance matrix (R) between the data was calculated using Eq. 3, where n and y̅ijT represent the wavenumber and the transposed standardized matrix y̅ij.

$$R=cov\left({{\widehat{y}}_{ij}}^{T},{\widehat{y}}_{ij}\right)=\frac{1}{n-1}\left({{\widehat{y}}_{ij}}^{T}\bullet {\widehat{y}}_{ij}\right)$$
(3)

After that, the eigenvector (EIJ) and eigenvalues (λI) were determined using Eq. 4, where the EIJ represented the degree of importance of specific wavenumbers corresponding to the specific PCs.

$${R}_{IJ}{E}_{IJ}={\lambda }_{I}{E}_{IJ}$$
(4)

A threshold of 60% total cumulative variance (CVAR) was set to decrease the variable size I number of PCs to ensure minimal information is lost. Consequently, the variance was calculated by the eigenvalues (λN) of respective PCs divided by the sum of all eigenvalues (Σλ).

$$CVAR={\sum }_{I=1}^{I}{VAR}_{I}\ge 60\mathrm{\%}$$
(5)
$${VAR}_{I}=\frac{{\lambda }_{I}}{\sum \lambda }$$
(6)

The specific PCs score was calculated with the dot product between the initial standardized data (ŷij) and the calculated eigenvector (EIJ). With this, the eigenvector represented the direction of PCs score of the specific wavenumber.

$${F}_{iI}={\widehat{y}}_{ij}\bullet {E}_{IJ}$$
(7)

Hence, the generated PCA graph of the honey samples with wavenumber showed the data visualization and allowed for the inspection of the grouping algorithm.

Results and Discussion

Melissopalynological Analysis

The melissopalynological analysis was conducted on 20 honey samples to identify the botanical origin of foraged nectar. Table 1 summarizes the identified 36 honey pollens from the 20 honey samples based on their characteristics described in several literatures (Rosdi et al. 2016; Yang et al. 2018) as well as the image of 21 unidentified pollens. The analysis on the frequency of pollens identified revealed that all samples were multifloral, typically collected from fruit plants, flowering plants and vegetables crops (Table 2).

Table 1 List of pollens found in 20 honey samples
Full size table
Table 2 Frequency of known and unknown pollens in 20 honey samples according to the botanical sources
Full size table

All the identification of 36 honey pollens indicates the availability of numerous nectar sources for bees, which may enhance the nutritional value of the honey. This finding is consistent with previous studies highlighting the significance of a varied range of nectar sources in honey production (Majid et al. 2020; Raja et al. 2014). The presence of multifloral honey in all samples suggests that bees in the study area foraged from a variety of plant species, which is beneficial for both the bees and the quality of the honey produced.

Manuka honey is expected to contain predominantly pollens from the manuka tree, Leptospermum scoparium. Interestingly, our study revealed that Apis Manuka honey samples also contain high amount of Tridax procumbens Linn or coatbuttons or Tridax daisy pollens. T. procumbens is a common weed but is also used as an Ayurvedic herb of Asia with a history of traditional use, particularly for wound healing (Ingole et al. 2022).

Incidentally, both Acacia and Manuka Apis honey samples generally exhibited lower total pollen (6–9) while those found in SBH are more diverse, ranging from 4 to 15 pollens. More types of pollens in some SBH samples may be attributed to the fact that stingless bees have smaller bodies than Apis bees, which enables them reach to wider range of botanical sources, as they can easily penetrate the deepest part of flowers (Selvaraju et al. 2019). Apis bees are unable to do so due to their larger body size.

It was observed that both Apis Manuka honey samples (Manuka1 and Manuka2) contained similar pollen, despite being collected from different regions (Australia and New Zealand). This suggests that both Apis Manuka bees have similar preferences for botanical sources when collecting nectars, where Leptospermum scoparium (Manuka) and Tridax procumbents (Tridax daisy) are among the most widespread weeds and perennial scrubs available in Australia and New Zealand (Thalkari et al. 2020). This finding may be useful in identifying the origin of Manuka honey samples from different regions.

Acacia mangium (mangium) and Elaeis guineensis (oil palm) were identified as predominant pollen in both Apis Acacia honey samples from Sibu, Sarawak. This suggests the existence of a wide plantation of E. guineensis around the bee farms within 5 to 9 km, as Apis bees with larger body size can travel up to 5 to 9 km for foraging (Beekman and Ratnieks 2000). This finding could be useful in developing strategies to promote the sustainable production of honey in the region. Additionally, A. mangium is widely used in industrial plantations in Sarawak as a commercial source of logs (Lee 2018), which can exude floral nectar and attract Apis bees to collect floral nectar throughout the year. Hence, it is likely that the beekeeping farms were located around A. mangium plantation, serving as the botanical sources of the honey samples and consequently minimizing the cost of honey sources plantation. The predominant pollen of SBHP2 was identical to the botanical source claimed by the producer, A. mangium, supporting the accuracy of the melissopalynological analysis conducted in this study. This finding highlights the importance of using reliable methods for pollen analysis in apicultural. As a result, the pollen analysis can illustrate the possible botanical sources that contribute to the honey properties. The most common plant pollen found in raw SBH is T. procumbens, which is a widespread and perennial herb found in tropical and subtropical regions. With this, it is readily available as a common weed, which appears to be the predominant pollen in several honey samples (SBHR2, SBHR3, SBHR7, SBHR8, SBHR9, SBHR10, SBHR11 and SBHR12). Further study on raw SBH with T. procumbens as the predominant pollen can be conducted as it has several potential therapeutic effects, such as wound healing, exhibiting antimicrobial, anti-inflammatory and wound healing properties (Ingole et al. 2022; Thalkari et al. 2020).

Most of the raw SBH samples presented several types of pollen from fruit trees, vegetables, cereal grains and seasoning species. The common fruit tree pollens were Musa acuminata (banana), Cocos nucifera (coconut), Durio zibethinus (durian), Dimocarpus longan (longan), Litchi chinensis (lychee) and Garcinia hombroniana (mangosteen). In addition, the common vegetables were Allium cepa (onion), Cucumis sativus (cucumber), Momordica charantia (bitter gourd), Solanum lycopersicum (tomato) and Solanum melongena (eggplant). The common seasoning and spices were Capsicum annuum (chilli) and Cinnamomum cassia (cinnamon), while Zea mays (corn) was the common cereal grain found in the honey samples. It is worth noting that all the pollen of most tropical crops have smaller-sized pollen flowers, which are easily pollinated by the stingless bees with smaller body sizes than Apis bees with bigger body sizes (Abd Jalil et al. 2017). As a result, the smaller-sized pollens found in the honey samples could act as the most significant markers to differentiate Malaysian honey from imported honeys (e.g. Australia honey and New Zealand honey), as the botanical sources (Durio zibethinus, Cocos nucifera) of SBH are locally grown and obtained from tropical countries.

Having the information on the common botanical source of the honey can also shed light to the potential nutritional and health benefit potential of the honey sample. For example, two raw SBH samples (SBHR4 and SBHR5) were found to be derived from Cinnamomum cassia (cinnamon), Cocos nucifera (coconut) and Musa acuminata (banana). Previous research conducted on Cinnamomum cassia showed it can decrease oxidative stress in a dose-dependent way through the inhibition of 5-lipoxygenase (Dugoua et al. 2007), with the presence of antioxidant phytochemicals (e.g. camphene, epicatechin, phenol and tannins (Hamidpour et al. 2015)). Another prior study conducted on Musa acuminata showed promising 2,2-diphenyl-1-picrylhydrazyl (DPPH) scavenging potential with low IC50 (7.63 mg/mL) in the methanolic extract of Musa acuminata flower (Sumathy et al. 2011) and much lower IC50 (0.19 mg/mL) in the methanolic extract of Musa acuminata flower stalk (Lee et al. 2011). Moreover, another research indicated that the hydrosol produced by coconut flowers showed the highest total phenolic content compared to the coconut fruit and leaves (Jakubczyk et al. 2021), which showed that the coconut flower pollen collected by the bees could contribute to the honeys’ phenolic composition. As a result, the physicochemical properties of the botanical sources in honey samples were influential to the honey composition such as phenolic composition and antioxidant activities.

Attenuated Total Reflection Fourier-Transform Infrared (ATR-FTIR) Analysis

The chemical composition of honey is a crucial factor for determining its quality and nutritional value. In this study, we employed ATR-FTIR analysis, a simple, non-destructive and reagent-free method, to determine the chemical structure of various honey samples. The analysis revealed that the honey samples had varying carboxylic acid contents, which was influenced by their botanical origin, geographical location and environmental pollution (Kędzierska-Matysek et al. 2018), where the peak signals in the FTIR spectra were dependent on the components present in the honey (Table 3). Figure 1 shows a sample of infrared spectra obtained from the honey samples while the presence and absence of the peaks at selected frequency regions, with their assigned vibrations, are summarized in Table 3. All show similar peaks, but slight differences in absorbance values, suggesting different concentrations of respective chemical component for different honey samples. Among all the honey samples, the C-O and C–C stretching vibrations exhibited the highest absorbance values, displaying peaks in the range of 1000 to 1200 cm−1. This can be attributed to the chemical component of sucrose, glucose, and fructose, responsible for the sweetness of the honey samples (Ismail et al. 2021). In contrast, the lowest absorbance values were obtained at 2850 to 3000 cm−1, corresponding to C-H stretching vibrations from cellulose and lipids.

Table 3 FTIR analysis of honey samples
Full size table
Fig. 1
figure 1

A sample of FTIR spectra of honey

Full size image

Fingerprint Region (450 to 1000 cm− 1)

The fingerprint region between 450 and 1000 cm−1 contains numerous infrared signals that can be used to identify the molecular composition of a sample (in this case, honey samples). However, this region is also complex, making it challenging to interpret, and for these reasons, this region is rarely discussed (Ramírez-Hernández et al. 2019). The PCA scores plot illustrated in Fig. 2 shows that the 20 honey samples could be clearly differentiated into four distinct groups. Therefore, chemical compounds from the botanical origins could contribute to the ‘fingerprint region’. Moreover, SBHR8 showed a similar fingerprinting region with two processed SBH from West Malaysia (SBHP1 and SBHP2), while the other two processed SBH from Sarawak (SBHP3 and SBHP4) showed similar fingerprinting region to most of the raw SBH. Interestingly, three raw SBH samples (SBHR3, SBHR6 and SBHR12) did not cluster together with other raw SBH samples, indicating that they may have different chemical compositions. Future studies should focus on identifying the chemical compounds responsible for these differences.

Fig. 2
figure 2

Principal component analysis of the fingerprinting region (450 to 1000 cm−1) in all the honey samples

Full size image

Stretching Vibrations from Water (3250 cm− 1)

The FTIR spectra of all honey samples showed a distinct peak at 3250 cm−1, indicating the presence of O–H stretching vibrations from water (Tahir et al. 2017). It is worth noting that O–H stretching vibrations from water in honey can also be affected by other factors such as temperature, humidity and storage conditions. In addition, the presence of other chemical groups such as carbohydrates, proteins and organic acids in honey could also contribute to the overall FTIR spectrum.

Honey contains a significant amount of water content, ranging from 18.22 to 36.82% (James et al. 2009), which could explain the high peak signal achieved in all honey samples (Table 3). Of these samples, Manuka2 and SBHR6 showed higher transmittance values when compared to other samples (~ 7% higher), suggesting the higher presence of hydroxyl bond (-OH) from phenolic content, alcohol, and water formation in these honey samples. The O–H stretching vibrations from water are mainly dependent on the hydrogen bonds that occur in dimeric carboxylic acids (Tahir et al. 2017). These vibrations are also affected by the presence of other chemical groups that may contribute to the formation of hydrogen bonds, such as phenolic compounds and alcohols. Therefore, the differences in the transmittance values of the O–H stretching vibrations from water observed in Manuka2 and SBHR6 could be attributed to the presence of these compounds.

Vibrations in Lipids

The vibration at wavenumber of 1740 cm−1 is depending on the C = O stretching vibrations from lipids, and it will be affected by the presence of rape pollen in the honey (Kasprzyk et al. 2018). Although lipids are present in small quantities in honey, it has been discovered to comprise various neutral lipids such as cholesterol esters, fatty acids, hydrocarbons, sterols, waxes and esters of polyols (Gholami et al. 2022).

Based on the results (Table 3), several raw SBH (SBHR1, SBHR2, SBHR6, SBHR8, SBHR9, SBHR10, SBHR11), processed SBH (SBHP3, SBHP4), processed ABH (ABHP1, ABHP2) and Manuka honey (Manuka1, Manuka2) did not display any peak signals at the wavenumber of 1740 cm−1, suggesting that the rape pollen was not present in the honey samples. This is because of the low-frequency stretching vibrations of carbonyl group (C = O) of the ester bond presented in lipid-based bee pollen constituents (Socrates 2004).

As for the remaining honey samples, two processed SBH samples (SBHP1, SBHP2) showed higher peak signals than the raw SBH samples (SBHR3, SBHR4, SBHR5, SBHR7, SBHR12), demonstrating a higher presence of locally sourced plant pollen with high lipid-based contents in the honey samples. This is because the locally sourced plant pollen membranes normally contain lipids, which include choline, glycerol, inositol and serine (Dahl 2018), and can eventually affect the vibrations of lipids in the honey samples.

Vibration in Protein (N–H Deformation and C–N Stretching Vibrations from Amide III – 1200 to 1350 cm−1; N–H Deformation and C–N Stretching Vibrations from Amide II – 1540 cm−1; C = O Stretching Vibrations from Amide I—1650 cm−1; N–H Stretching Vibrations from Proteins – 3300 to 3500 cm−1)

Protein is a minor component in honey, and it can be used as a marker for taxonomic classification in honey. The present FTIR analysis successfully identified the protein peptide bond (an amide group) based on specific wavenumber ranges: amide I at 1600 to 1800 cm−1, amide II at 1470 to 1570 cm−1, amide III at 1250 to 1350 cm−1 and amide A at 3300 to 3500 cm−1 (Ji et al. 2020). All the honey samples showed similar peak signals in 1200 to 1350 cm−1 and 1650 cm−1, where a shoulder peak at 1200 to 1350 cm−1 was based on the amide III that related to the in-phase combination of N–H deformation and C-N stretching vibrations. Besides, the bands appearing in 1600 to 1800 cm−1 (mainly 1650 cm−1) are mainly related to phenolic molecules, with the result of stretching in the carboxyl group (C = O and C≡C) (Tahir et al. 2017).

All the honey samples showed a weak absorption at 1540 cm−1, which is related to the deformation of N–H group and stretching vibrations of C-N from amide II. As a result, a weak and broad peak of N–H stretching vibrations from proteins at 3300 to 3500 cm−1 was observed in all the honey samples due to the small amount of protein content in the honey samples. This is because the protein content in the honey mainly depends on the presence of proteolytic enzymes (Rossano et al. 2012) and the dietary proteins in the bees are principally digested by three different midgut endopeptidases, which include chymotrypsin, elastase and trypsin, as well as exopeptidase leucine aminopeptidase (Burgess et al. 1996).

Principal component analysis (PCA) was conducted between the selected wavenumber (1600 to 1800 cm−1, 1470 to 1570 cm−1, 1250 to 1350 cm−1, 3300 to 3500 cm−1) and total protein content. According to Fig. 3, most of the honey samples (except Manuka1, Manuka2, ApisP2, SBHR3, SBHR6 and SBHR12) were clustered together due to the similar amount of N–H, C-N, C = O and total protein content found in the honey samples. Hence, the amount of protein (e.g. albumins, globulin, peptones some nucleo-protein) occurred in the remaining honey samples could be slightly different than the honey samples. Therefore, specific analyze of protein extraction in honey should be carried out in future study.

Fig. 3
figure 3

Principal component analysis of the transmittance of 1200 to 1350 cm−1, 1540 cm−1, 1650 cm−1, 3300 to 3500 cm−1 and total protein content in all the honey samples

Full size image

Conclusions

The honey samples were evaluated as multifloral honey due to several pollen sources presented in the honey samples. As a result, the honey samples were considered pure and unadulterated honey due to the presence of diverse pollen sources. There were 36 different pollen sources recognized in this study, which mainly consist of fruit plants, flowering plants and vegetables. The botanical sources with high antioxidant properties and phenolic composition in several SBH affirmed the contribution of the potential of antioxidant properties. All in all, melissopalynological analysis serves as an efficient basic tool in identifying the botanical origins of honey samples. While more advanced techniques, such as DNA barcoding, can provide further validation of pollen origins, they need more expensive equipment, specific chemicals and special skills. So, the simpler analysis method was still very effective in understanding where the honey comes from.

The results of ATR-FTIR spectroscopy illustrated the spectral patterns of all the honey samples were nearly identical but significant differences in the absorbance peak and basic analysis were observed. However, more studies on physicochemical qualities are recommended for a more complete analysis and better discrimination. This is because the chemometrics approach is mainly used as an assumption of analyzing the samples’ backgrounds through the current statistics.

Hence, the outcome of this study is essential in providing valuable insights into establishing the quality standard of stingless bee honey and has the potential to improve the present Malaysian Standard MS2683:2017 for further commercialization purposes.