Mediterranean Propolis from the Adriatic Sea Islands as a Source of Natural Antioxidants: Comprehensive Chemical Biodiversity Determined by GC-MS, FTIR-ATR, UHPLC-DAD-QqTOF-MS, DPPH and FRAP Assay.

There is no systematic report about propolis chemical biodiversity from the Adriatic Sea islands affecting its antioxidant capacity. Therefore, the samples from the islands Krk, Rab, Pag, Biševo and Korčula were collected. Comprehensive methods were used to unlock their chemical biodiversity: headspace solid-phase microextraction (HS-SPME) and hydrodistillation (HD) followed by gas chromatography and mass spectrometry (GC-MS); Fourier transform mid-infrared spectroscopy (FT-MIR); ultra high performance liquid chromatography with diode array detector and quadrupole time-of-flight mass spectrometry (UHPLC-DAD-QqTOF-MS) and DPPH and FRAP assay. The volatiles variability enabled differentiation of the samples in 2 groups of Mediterranean propolis: non-poplar type (dominated by α-pinene) and polar type (characterized by cadinane type sesquiterpenes). Spectral variations (FT-MIR) associated with phenolics and other balsam-related components were significant among the samples. The UHPLC profiles allowed to track compounds related to the different botanical sources such as poplar (pinobanksin esters, esters and glycerides of phenolic acids, including prenyl derivatives), coniferous trees (labdane, abietane diterpenes) and Cistus spp. (clerodane and labdane diterpenes, methylated myricetin derivatives). The antioxidant potential determined by DPPH ranged 2.6–81.6 mg GAE/g and in FRAP assay 0.1–0.8 mmol Fe2+/g. The highest activity was observed for the samples of Populus spp. origin. The antioxidant potential and phenolic/flavonoid content was positively, significantly correlated.


Introduction
Apis mellifera L. propolis, known as the bee glue, combines resins collected by the honey bees from different plant organs, and with beeswax that honey bees additionally incorporate. It has been shown that propolis possesses antioxidant, antibacterial, antifungal and antiviral properties, as well as 2.6. Total Flavonoid (TF), Total Phenolic (TP) Content and Antioxidant Potential (DPPH and FRAP Assays) 2.6.1. Total Antioxidant Activity (FRAP Assay) The ferric reducing antioxidant assay (FRAP) was performed as previously described [20,21]. Briefly, the reagent was prepared by mixing 10 mmol/L TPTZ reagent (2,4,6-tri(2-pyridyl)-s-triazine) with 20 mmol/L ferric chloride in acetate buffer (pH 3.6). The quantitative results were calculated using a calibration curve of ferrous sulfate used as external standard (0.02-1.5 µmol/mL). Before the analysis, the propolis extracts were diluted 20-200 times and 20 µL of the extract solutions were mixed with 200 µL of ferric complex. The results were calculated and expressed as micromoles of Fe 2+ per gram of propolis. The absorbance (λ = 593 nm) was read in disposable optical polystyrene 96-well plates (FL medical, Torreglia, Italy) using a Multiskan™ GO Microplate Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). All the measurements were performed in triplicate.

Total Phenolic Content (TP)
The total phenolic content was measured spectrophotometrically using the Folin-Ciocalteu method, as previously described [22,23]. Before the analysis, the propolis extracts were diluted 20-200 times and 50 µL of the ethanolic extract solution were mixed with 20 µL of Folin-Ciocalteu reagent. After 5 min, 200 µL of 100 g/L Na 2 CO 3 solution was added. After 90 min of incubation at room temperature, Antioxidants 2020, 9, 337 5 of 32 in dark, the absorbance was read against blank (prepared similarly, using pure solvent instead of sample) at 725 nm in disposable polystyrene 96-well plates using a microplate spectrophotometer. Total phenolic content was calculated using a calibration curve prepared with fresh gallic acid standard solutions (10-200 µg/mL) and expressed as milligrams of gallic acid equivalent (GAE) per gram of propolis. All the measurements were performed in triplicate.

Total Flavonoid Content (TF)
The total flavonoid content was measured spectrophotometrically using a modified pharmacopoeial method with aluminum chloride [24]. An aliquot of 50 µL of prepared extracts was mixed with 50 µL of 2% ethanolic solution of AlCl 3 (w/v) and after 60 min of incubation at room temperature, in dark, the absorbance was measured at 420 nm using a microplate reader. Total flavonoid content was calculated using a calibration curve prepared with quercetin standard solutions (20-400 µg/mL) and expressed as milligrams of quercetin equivalent (QE) per gram of propolis. All the measurements were performed in triplicate.

Antiradical Activity (DPPH Test)
Determination of antiradical activity using DPPH radical and comparison with the gallic acid activity was performed using previously modified method [25]. Before the analysis, the propolis extracts were diluted 20-200 times and 20 µL of diluted test extracts were mixed with 200 µL of 0.315 mM DPPH solution in methanol and incubated for 30 min at room temperature, in dark. The absorbance (λ = 517 nm) was read in disposable optical polystyrene 96-well plates using microplate spectrophotometer (as previously). All the measurements were performed in triplicate. The obtained data were calculated from appropriate gallic acid calibration curve (2.0-100 µg/mL) and expressed as gallic acid equivalent antioxidant capacity per gram of propolis (mg GAE/g).

Statistical Analysis
Statistical analysis was performed for correlation of the antioxidant potential and phenolic/flavonoid content using STATISTICA 64 ver. 13.1 (Dell Inc., Tulsa, OK, USA). Pearson's product-moment correlation was applied to test relations between the investigated parameters and significance was assessed in two-tailed test at the level of significance p < 0.05.

HS-SPME/GC-MS and HD/GC-MS
HS-SPME has been used in last decade for the analysis of propolis headspace (HS) volatile organic compounds (VOCs) as a simple and fast method. To obtain comprehensive HS chemical profiles among samples 3 types of fibers were used. For the isolation of volatile and less-volatile compounds HD with solvent trap was used. VOCs composition is strongly dependent on the extraction method. Striking differences were found between chemical profiles of the same sample obtained by HS-SPME and HD and among the samples. It is known that to produce propolis, bees collect various exudates including balsams, resins and waxes from the plants available in specific areas. It results in different typologies of the final product and therefore the samples were divided (according to VOC results) into two groups depending on the probable plant sources.

Mediterranean Propolis (Non-Poplar Type)
According to the chemical composition of HS and essential oil (EO) (Tables 1 and 2), the samples BP and KP were classified in this group. Those two samples were found peculiar as expected, since the islands Biševo and Korčula are more distant from the mainland and are populated by a specific flora. Table 1. Volatiles determined by headspace solid-phase microextraction (HS-SPME)/gas chromatography (GC-MS).

No.
Compound RI  BP  KP  RP  PP  K1P  K2P  K3P  I  II  III  I  II  III  I  II  III  I  II  III  I  II  III  I  II  III  I  II  III   1 Ethanol Acetone < 900 6.8 5.7 0.9 1.9 2.6 1.

Comparison with Probable Plant Source Volatiles
Cupressus spp. and Juniperus spp. have been most frequently reported in last decade as sources of Mediterranean type propolis [27,31] and those plants are naturally widespread, among others parts of the Adriatic region, on the islands Biševo and Korčula as reported in the Flora Croatica Database [36]. α-Pinene found in the investigated samples could originate from exudates of Cupressus sempervirens L. known as one of the source plants utilized by the bees to form propolis [27,31] and it is known that EO of C. sempervirens from Croatia [37] contains α-pinene as the main component (up to 79.2%). In propolis samples from Southern Italy (Adriatic coast) and Greece α-pinene was also identified at high percentage and other coniferous species were also suggested as the plant source [27,31]. However, the abundance of monoterpene fraction, with a high α-pinene content, was also described for the species of the genus Juniperus [38] and it was reported that monoterpenes may also contribute to propolis in specific geographical locations [34]. α-Pinene is found as the main component of the needles EO (41.37%) and berries EO (66.30%; 61,21%) of the wild Croatian Juniperus oxycedrus L. [39]. However, the contribution of other plant sources is also possible, especially Pinus spp. that are well known to contain α-pinene in the resin EO (21.39-25.40% [40]) and in the headspace (66.2%; 73.4% [41]). Limonene, found with minor abundance in VP HS and EO, was identified in the EO of C. sempervirens and J. oxycedrus from Croatia [37][38][39] and Pinus spp. resin [40,41]. Manoyl oxide (12.29%) and α-campholene aldehyde (0.15%) were present in J. oxycedrus needles EO from Croatia [39], in Pinus spp. resin (0.4-0.9% [41]) and in BP EO. Manoyl oxide was also identified in C. sempervirens EO [37,38]. Manool and guaiol present in KP EO were also found in Cupressus and Juniperus plants [42,43]. Tricyclic diterpenes (particularly methyl isopimarate, dehydroabietic acid, dehydroabietane, dehydroabietal) were found in Juniperus plants [43]. Abietane diterpenoids from C. sempervirens were also reported [44] and dehydroabietane was isolated from the cypress EO [45] and in BP EO. α-Cedrol, having a woody and spicy characteristic smell, was found in C. sempervirens EO at 23.68% [45] and in the range 1.2-12.9% [37]. Although it was not found in J. oxycedrus from Croatia [39] it was found in J. oxycedrus from Turkey (2.3%-9.7% [44]). Some of the volatile compounds in BP EO may be linked to other Cistus species-for example Cistus salvifolius L. and their chemotypes, that provide high volatiles diversity and are dominated by oxygenated sesquiterpenes and monoterpenes [46]. Manoyl oxide is also one of the major constituents of BP EO, but was also the main component of essential oil obtained from Cistus creticus L. [47], while essential oil of C. creticus subsp. eriocephalus was characterized by i.a. manoyl oxide, αand δ-cadinene, viridiflorol and bulnesol [48]. On the other hand, essential oil of other Cistus species cultivated in Corsica, such as Cistus ladaniferus L. was dominated by α-pinene (11.1-47.4%), that is another relevant compound found in BP EO [49]. In general, the size of n-alkanes from black pine needles wax ranged from C 16 to C 33 and the most abundant were C 23 , C 25 , C 27 and C 29 [50]. Cupressaceae leaf wax has been characterized (chemotaxonomic significance) by moderate percentages of n-alkanes [51], particularly of C 31 , C 33 , C 27 and C 21 , including C 22 , C 23 and C 24 . Heneicosane, docosane, tricosane and tetracosane were found as major constituents in KP EO. Heneicosane was found with very high percentages only in C. sempervirens [52]. In addition, all leaf-wax samples of J. communis showed predominance of n-alkane C 33 in the needle wax (30.0-61.4%), which appears to be a common feature for Juniperus species [53] (the range of n-alkanes reported by different authors varied from mid-length (C 23 ) to long-chain n-alkanes (C 25 -C 35 )).

Mediterranean Propolis (Poplar Type) Volatiles
According to the chemical composition (Tables 1 and 2), the samples RP, PP, K1P, K2P and K3P were classified in this group. The islands Rab, Pag and Krk are located closer to the Adriatic coast, characterized by an abundance of Populus spp. [36].

Comparison with Populus spp. Volatiles
Poplar spp. (Populus nigra L., Populus tremula L. and Populus alba L.) and the buds resin have been reported as a primary source of propolis from temperate zones [6]. As reported in the Flora Croatica Database [36], the area of the islands Pag, Rab and Krk is within the range of different Poplar ssp. abundance (particularly P. nigra, P. tremula and P. alba). Black poplar (P. nigra) buds exhibited different EO profiles (both qualitatively and quantitatively). Some buds contained mainly oxygenated sesquiterpenes, particularly α-, βand γ-eudesmols that are present in P. nigra buds EO [60,61] as well as their CO 2 extracts [20]. These compounds are present in K1P, K2P and K3P EO, PP HS and EO as well as RP HS and EO that could be connected with the P. nigra (eudesmol chemotype) distribution on the islands Krk, Rab and Pag. P. nigra buds EO [20] contained bulnesol (4.4%) and guaiol (5.7%), as well as their supercritical CO 2 extracts (guaiol (2.7-3.7%); bulnesol (2.5-3.4%)). Hexane extracts of P. nigra buds [62] analyzed by GC-MS contained, among other constituents, guaiol (8.7%) and bulnesol (3.8%). Isomers guaiol and bulnesol were characteristic for PP HS and EO as well as RP EO indicating dominant influence of P. nigra (bulnesol/guaiol chemotype) from the islands Pag and Rab. Several P. nigra buds EO [60] were mainly composed of sesquiterpene hydrocarbons (mainly ar-and γ-curcumene and δ-cadinene). Isomers δ-, γand α-cadinene were typical for RP HS and EO that can be also connected with P. nigra (cadinene chemotype) distribution on the island Rab. P. nigra buds EO was reported also with different profiles containing a mix of sesquiterpenes and derivates of benzoic acid, mainly prenyl benzoate [60]. About 50% of the GC chromatograms of hexane extracts of P. nigra buds [62] consisted of higher alkanes including docosane and tricosane (found in RP, PP, K1P, K2P and K3P EO), but C 25 -C 31 alkanes dominated (they were not present in investigated propolis EO probably due to lower volatility). Higher alkanes are known to be one of the main components of cuticular waxes of plant leaves and stems. Aliphatic alcohols 2-methylbut-3-en-2-ol, 3-methylbutan-1-ol, (E)-2-methylbut-2-en-1-ol and (E)-2-metylbut-2-enoic acid were identified in P. nigra buds EO [61]. These hemiterpenes were present in total amount up to 8%. Hemiterpenes were present in RP HS, PP HS, K1P HS and K3P HS. The esters of hemiterpene (prenyl) alcohols and cis/trans caffeic, ferulic and isoferulic acids were previously identified in the bud exudate of P. nigra [63], but as non-volatile compounds they cannot be isolated by hydrodistillation. Aspen buds (P. tremula) also exhibited different EO profiles. Several aspen buds [60] contained mostly benzoic acid derivates (benzyl benzoate, salicyl benzoate and trans-benzyl cinnamate). They were identified in K2P EO indicating aspen (P. tremula)-type propolis and P. tremula is reported on the island Krk [36].

Chemical Characterization by FTIR-ATR Spectroscopy
In most of the FTIR spectroscopic studies, propolis research has been focused on ethanolic propolis extracts (EPE) [64][65][66][67][68][69] while raw beehive propolis that serves as a source (raw material) for preparing propolis-based products (such as the most commonly used propolis ethanolic tincture) has been covered only by two reports [70,71].
Complexity of FTIR-ATR spectrum of raw propolis arises from its complex chemical composition that varies significantly depending on the source of the plant exudate which bees have collected. Still, chemical composition of propolis has generally been represented by two groups of constituents: balsam content (40-70%) mostly comprised of numerous phenolics, and non-balsam content containing beeswax (20-35%), essential oil (3-5%; mono and sesquiterpenes) and other organic compounds (ca. 5%; ash content, polysaccharides: proteins, amino acids, mechanical impurities, etc.). Balsam content is the most complex compositional segment of propolis and includes the following substances: phenols, phenolic acids, esters, flavanones, dihydroflavanones, flavons, flavonols, chalkones, phenolic glycerides and other minor compounds [72].
Given that FTIR spectrum of propolis reflects its overall chemical composition, identification of absorption bands, i.e., assignment of functional groups within the IR spectrum of raw propolis material represents a demanding task due to a large number of various organic compounds and corresponding molecular vibrations that can be observed in it. Nevertheless, it is possible to distinguish signals that are highly specific for particular organic compound based on the comprehensive literature data on propolis chemical composition, as well as various sources of FTIR spectral data (e.g., spectral libraries and atlases).
General assignment of molecular vibrations in the propolis spectrum is presented on an average FTIR-ATR spectrum of K3 sample from Krk island (Figure 1). The complexity of its absorptions is arising from a complex composition dominated by substances from the balsam group of compounds. A broad strong band at 3350 cm −1 observed in analyzed propolis samples occurs due to the O-H stretching vibration of the phenolic group. Spectral features related to phenols are also characterized by interaction of O-H deformation and C-O stretching vibrations which can be observed in the spectral range between 1405 and 1220 cm −1 (with maximum absorbance at 1375 cm −1 ) and in the form of series of weak vibrations between 1260-1180 cm −1 . Phenols are also represented with a doublet at 1640 cm −1 assigned to aromatic ring C=C stretching and aromatic C-H deformation vibration at 1110 cm −1 [73]. A medium absorption at 720 cm −1 is peaking due to CH 2 rocking of hydrocarbons originating from beeswax [72]. An overlapping effect with out-of-plane deformation of the O-H group of phenols is possible in this region. A weak band peaking at 1515 cm −1 can be assigned to flavonoids; C=C (aromatic ring) stretching [68]. C-H deformations and aromatic stretching at 1461 cm −1 is assigned to flavonoids (hydrocarbons CH 3 and CH 2 vibrations are overlapping). The most prominent absorption in the fingerprint region is a broad band with absorption maximum observed at 1170 cm −1 that corresponds to the C-O asymmetric stretching vibration of esters related to long-chain aliphatic acids. Saturated aliphatic esters typically absorb at 1750-1725 cm −1 [73]. Thus, absorption occurring at 1736 cm −1 is due to the carbonyl group (C=O) stretching vibrations of the ester bond. This vibration can be attributed to the monoesters from beeswax in propolis, as the major ester component of beeswax (~40%) [74]. As shown in Figure 1B, other medium and weak intensity absorption bands are attributed to the vibrations of various functional groups of phenols, flavonoids and hydrocarbons, some of which overlap.
flavonoids; C=C (aromatic ring) stretching [68]. C-H deformations and aromatic stretching at 1461 cm -1 is assigned to flavonoids (hydrocarbons CH3 and CH2 vibrations are overlapping). The most prominent absorption in the fingerprint region is a broad band with absorption maximum observed at 1170 cm −1 that corresponds to the C-O asymmetric stretching vibration of esters related to longchain aliphatic acids. Saturated aliphatic esters typically absorb at 1750-1725 cm −1 [73]. Thus, absorption occurring at 1736 cm −1 is due to the carbonyl group (C=O) stretching vibrations of the ester bond. This vibration can be attributed to the monoesters from beeswax in propolis, as the major ester component of beeswax (~40%) [74]. As shown in Figure 1B As presented in Figure 2, unique spectral patterns of propolis from different Adriatic Sea islands reflect compositional differences (different band positions and intensities) between the samples and indicate significant compositional differences. Variations in hydrocarbon content (at 2916, 2848, 1461, 730 and 720 cm −1 ) and esters (at 1736 cm −1 ) originating from beeswax present in propolis [70,73] are not distinguished significantly between analyzed propolis samples, as opposed to spectral variations associated with phenolics and other balsam-related components that are clearly observable. These differences are mainly related to the content of phenols, flavonoids and esters, and corresponding spectral variations are most prominent in the fingerprint region (1800-600 cm −1 ). The results of spectral analysis revealed great similarity of propolis samples from the islands Biševo and Korčula indicating similar botanical origin. Two propolis samples from Krk (K1P and K3P) were also found to be similar, while propolis from Pag, Rab, as well as K3P from Krk, showed specificities due to characteristic phenolic and ester bands (indicating that propolis was collected from different resin sources). As presented in Figures 3 and 4, fingerprint region displays a series of multiple absorption bands occurring due to mentioned groups of organic compounds. It can be observed that propolis from Biševo and Korčula exhibit similar spectral pattern in this region, while propolis from other islands (Pag, Rab, Krk) reflect unique spectral features. Among them, Pag propolis and K2P propolis (Krk propolis from Pinezići) are the most distinguished ones due to the high phenolic content (represented by the most prominent phenolic band at 1030 cm −1 ), while Rab (RP) and K2P propolis stand out for their higher ester content (absorption maximum at 1070 cm −1 ). As presented in Figure 2, unique spectral patterns of propolis from different Adriatic Sea islands reflect compositional differences (different band positions and intensities) between the samples and indicate significant compositional differences. Variations in hydrocarbon content (at 2916, 2848, 1461, 730 and 720 cm −1 ) and esters (at 1736 cm −1 ) originating from beeswax present in propolis [70,73] are not distinguished significantly between analyzed propolis samples, as opposed to spectral variations associated with phenolics and other balsam-related components that are clearly observable. These differences are mainly related to the content of phenols, flavonoids and esters, and corresponding spectral variations are most prominent in the fingerprint region (1800-600 cm −1 ). The results of spectral analysis revealed great similarity of propolis samples from the islands Biševo and Korčula indicating similar botanical origin. Two propolis samples from Krk (K1P and K3P) were also found to be similar, while propolis from Pag, Rab, as well as K3P from Krk, showed specificities due to characteristic phenolic and ester bands (indicating that propolis was collected from different resin sources). As presented in Figures 3 and 4, fingerprint region displays a series of multiple absorption bands occurring due to mentioned groups of organic compounds. It can be observed that propolis from Biševo and Korčula exhibit similar spectral pattern in this region, while propolis from other islands (Pag, Rab, Krk) reflect unique spectral features. Among them, Pag propolis and K2P propolis (Krk propolis from Pinezići) are the most distinguished ones due to the high phenolic content (represented by the most prominent phenolic band at 1030 cm −1 ), while Rab (RP) and K2P propolis stand out for their higher ester content (absorption maximum at 1070 cm −1 ).

UHPLC-DAD-QqTOF-MS
The ethanolic extracts of seven propolis samples were analyzed, disclosing high diversity between the samples collected from different Croatian islands. Nearly 120 compounds were identified or tentatively identified in the samples (mainly derivatives of phenolic acids, flavonoids and terpenes (Table 3). Selected major phenolics were quantified and significant differences in their abundance were found. Content of phenolics in RP was much higher than in other samples ( Table 4). The

UHPLC-DAD-QqTOF-MS
The ethanolic extracts of seven propolis samples were analyzed, disclosing high diversity between the samples collected from different Croatian islands. Nearly 120 compounds were identified or tentatively identified in the samples (mainly derivatives of phenolic acids, flavonoids and terpenes (Table 3). Selected major phenolics were quantified and significant differences in their abundance were found. Content of phenolics in RP was much higher than in other samples ( Table 4).
The major compounds reported in most samples of Croatian propolis were phenolic acids (ferulic, p-coumaric acid) and flavonoids (galangin, pinocembrin, chrysin) [16,18,[76][77][78]. These findings are very similar to the profiles of RP, K1P and K3P from the current study. However, Croatian samples that did not contain these compounds were also reported, demonstrating occurrence of different propolis types in Croatia [16,18]. Saftić et al. recently reported LC-MS analysis of propolis from different regions of Croatia, including Mediterranean samples. The latter contained diterpenes, e.g., pimaric acid, isocupressic acid (found also in some of the currently analyzed samples) but on the other hand also totarol, agathadiol and artepillin C (corresponding exact masses not found in the currently analyzed samples) [15]. These compounds were previously proposed as markers for Mediterranean propolis poor in flavonoids and phenolic acids, deriving mainly from Cupressus spp. [5].

Possible Botanical Origin of the Samples Based on LC-MS Profiles
All the samples contained at least traces of prenyl caffeates, recognized as typical for the most common European black poplar-type propolis [60], however their total amount was more relevant only in three samples RP, K1P and K3P (37.74, 11.49, 5.92 mg/g, respectively), demonstrating primary (RP) or secondary (K1P and K3P) contribution of P. nigra balsam in those specimens ( Table 4). The same samples contained also another typical compounds from this source, such as pinobanksin 3-O-acetate, chrysin, pinocembrin and others [60] which again were very abundant in RP. RP contained particularly high amount of isoferulic acid (8.30 mg/g), while K1P and K3P only its slightly elevated amount (0.67-0.80 mg/g); though the amount of isoferulic acid was 2-4 times higher than of ferulic acid, which suggests that the bud exudate was partially collected from P. nigra var. italica Münchh. [79]. Different profiles of other poplar polyphenols were described in literature [60,79]. The highest differences between various Populus species were connected to the presence of various flavonoids, presence of both flavonoids and phenolic acids as well as their monoesters, or the presence of mainly free phenolic acids. Sakuranetin was found in both P. nigra [60,80] and P. tremula [60]. This may partially explain the origin of higher amounts of this compound (4.45-16.36 mg/g) also in PP, K2P or even KP. The samples PP, BP and K2P, but partially also RP, K1P and K3P contained notable peaks of glycerides of phenolic acids that could be attributed to P. tremula: e.g., p-coumaroyl glycerol, acetyl-p-coumraoylglycerol, 2-acetyl-1,3-di-p-coumaroylglycerol, 2-acetyl-3-p-coumaroyl-1-feruloylglycerol [79]. However, various glycerides of phenolic acids were found also in other natural sources such as aerial parts of Asparagus officinalis L. [81] or Aegilops ovata L. [82].
Some samples (KP, PP, K2P, K3P and partially BP) contained also few peaks with pseudomoecular ion mass 453.3372 [M − H + ] − , that could be tentatively identified as oleanoic acid, moronic acid or masticadienonic acid and could be potentially attributed as deriving from Pistacia lentiscus L. resin [85]. Samples KP and BP (traces) contained also small amounts of cupressic and isocupressic acid [75], that could be linked to Cupressus spp. [86] however totarol or agathadiol (considered markers of this type of propolis) were not detected [5]. This may suggest other conifers as possible sources of this propolis. Nevertheless, literature data showed also variability of totarol concentrations in different samples obtained from Cupressus spp. [86].
Considering all the data, notable contribution of Poplar species as source of propolis from Rab, Pag, Krk is found in contrast to the samples from Korčula and Biševo, which is consistent to the geographical location of the islands and distribution of poplars in Croatia [36]. Among the samples, RP contained particularly high levels of compounds typical for black poplar bud exudate, which may indicate dominating contribution of poplar in RP. In other samples its amount was at least several-fold lower, suggesting secondary contribution in total mass of propolis. The samples RP, K1P and K3P contained higher contribution of isoferulic acid than ferulic acid, which indicated italica variety of P. nigra as source plant. On the other side, the occurrence of various glycerides along with sakuranetin present in all samples (except BP) may be connected to other Populus species, such as P. tremula. These sources are quite common for European propolis from temperate climate zone [79]. On the other hand, some of the samples contained different terpenoids and unusual flavonoids that could be attributed to Mediterranean plants. Abietic and pimaric acid isomers as well as cupressic and isocupressic acids may be attributed to different, common coniferous tree species such as Pinus, Cypressus and Juniperus. These sources were also indicated as possible for other Mediterranean propolis, containing diterpenes, from Croatia but also west Algeria and Crete [15,87,88]. These compounds were more abundant in BP and KP. Some of the samples contained other compounds, possible to link with Pistacia. This source was suggested as one of the possible for Moroccan propolis [89]. The most interesting propolis, collected on island Biševo, distant from the land, was much different from other samples and contained compounds that could be linked to Cistus spp. Such origin was suggested also for diterpene propolis from Algeria which was very similar to the investigated Croatian sample [75]. Similarly, Tunisian propolis containing myricetin 3,7,4 ,5 -methylether, that is typical for Cistus spp. leaf exudates, was recognized as source plant [90]. This confirms, that in areas where poplars are not always available, other plant sources can be used to form propolis.

Total Phenol Content and Antioxidant Potential
The total content of phenolic compounds assessed using Folin-Ciocalteu reagent ranged from 14.0 to 189.7 mg GAE/g of propolis for the investigated samples. The highest value was observed for RP and the lowest for BP, however all other values did not exceed 37 mg GAE/g. The total flavonoid content ranged between 7.2 and 103.9 mg QE/g of propolis, and was the highest in RP and lowest in K2P, but it did not exceed 18 mg QE/g for any other samples. These result are consistent with those obtained by UHPLC, where content of phenolics in RP was much higher than in other samples ( Table 4). The values obtained for RP, were very similar to those reported for Chinese poplar propolis (233.98 mg GAE/g, 124.92 mg QE/g) and extracts from poplar buds (145.54 mg GAE/g, 126.23 mg QE/g) [123]. These values are consistent also with other obtained from other samples of Chinese poplar-type propolis that ranged from 87.11 to 257.93 mg GAE/g and 105.25 to 351.25 mg QE/g as well as those obtained for Croatian propolis 70-220 mg GAE/g [16,96]. This may suggest, that this sample was mostly originating form Populus exudates, while the other samples may contain no more than just a small percentage of this balsam. Similar observation was done for Anatolian propolis, where 3 different types were identified including those deriving from P. nigra, P. tremula and non-poplar type propolis. The amount of phenolics and flavonoids in the latter two, ranged from 11.24 to 47.15 mg GAE/g and from 3.88 to 48.70 mg QE/g, respectively [124]. Interestingly, the non-poplar type propolis from Anatolia was found to contain mainly Pinaceae and Cistus spp. pollen which suggest such plants to be major sources of these samples [124].
The antioxidant potential determined by DPPH ranged 2.6-81.6 mg GAE/g and in FRAP assay 0.1-0.8 mmol Fe 2+ /g. The highest activity was observed for RP and the lowest in BP. The antioxidant potential and phenolic/flavonoid content were positively (Table 5), significantly correlated (TP-DPPH R 2 = 0.9368, TP-FRAP R 2 = 0.7870, TF-DPPH R 2 = 0.9019, TF-FRAP R 2 = 0.7060, at p < 0.05) which links the activity with these groups of compounds. More varied activities in FRAP test ranging from 0.04 to 1.3 mmol Fe 2+ /g were found for Croatian propolis by Tlak-Gajger et al. [16].
Comparison of the obtained results with other reports on Croatian propolis was not possible, due to the different extraction, methodology or way of data presentation [18,78,125,126].

Conclusions
Typical propolis from Croatian islands along Adriatic Sea coast (Krk, Rab, Pag, Biševo and Korčula) were collected. The volatiles of the samples were isolated by HS-SPME and HD followed by GC-MS. The variability of the volatiles enabled differentiation of the samples in 2 groups of Mediterranean propolis: non-poplar type (dominated by α-pinene) and poplar type (cadinane type sesquiterpenes). Spectral variations (FT-MIR) associated with phenolics and other balsam-related components were significant among the samples. The quantitative data obtained from colorimetric tests and UHPLC-DAD suggests that only one sample was a typical black poplar-type propolis (characterized e.g., by abundance of caffeic acid prenyl esters, pinobanksin-3-O-acetate, pinocembrin). Few samples contained just its small, but visible contribution and derive mostly from other botanical sources such as other poplars or coniferous trees (e.g., Pinus, Cupressus or Juniperus). The latter may be linked with presence of abietic, dehydroabietic or pimaric acids, 6 -O-p-coumaroyltrifolin. One sample from Biševo was most particular and could be classified as Mediterranean diterpene propolis that derived i.a. from Cistus spp. exudates (characteristic compounds included myricetin-3,7,4 ,5 -tetramethyl-ether, 15-hydroxy-cis-clerodan-3-ene-18-oic acid, 18-hydroxy-cis-clerodan-3-ene-15-oic acid, 18-acetoxy-cis-clerodan-3-ene-15-oic acid). The highest activity was observed for the samples of Populus origin. The antioxidant potential and phenolic/flavonoid content was positively, significantly correlated.